SECTOR REPORT ASIA TECHNOLOGY - BNP Paribas · 2016-09-30 · Autonomous driving is a goal for...

錯誤 ! 使用 [常用 ] 索引標籤將 Title 套用到您想要在此處顯示的文字。 BNP PARIBAS Jimmy H uang

THIS REPORT IS SOLELY FOR RECIPIENTS WHO ARE NON-CHINESE INVESTORS LOCATED OUTSIDE THE PEOPLE'S REPUBLIC OF CHINA ("PRC"). WITHOUT THE PRIOR WRITTEN CONSENT OF BNP PARIBAS, THE RECIPIENTS OF THIS REPORT SHALL NOT FURTHER DISTRIBUTE SUCH REPORT OR DISCLOSE ANY INFORMATION THEREIN TO ANY OTHER PERSON INCLUDING BUT NOT LIMITED TO ANY OTHER NON-CHINESE INVESTORS OR ANY PERSON IN PRC. 30 SEPTEMBER 2016 SECTOR REPORT

ASIA TECHNOLOGY

Autonomous vehicles: a megatrend in the making

Autonomous driving is set to become a megatrend in China Leading technology giants and automakers have set their sights on autonomous driving, with

aggressive investments from the likes of Apple, Google, Mercedes Benz, BMW and Tesla. Although policies covering autonomous driving in China lag those in the US and EU, this is likely to change soon and more favourable policies could be a near-term driver. We expect autonomous vehicles to be a megatrend in China, with Baidu being a leading player; it targets mass production in 2021.

Five technologies that should benefit

Autonomous vehicles rely on advanced driving assistant systems (ADAS) and connected car systems. Both rely on five critical technologies: 1) sensors or radars, 2) deep learning (AI), 3) high definition mapping, 4) human machine interfaces (HMI), and 5) motion control. In related Greater China supply chains, international vendors dominate, but we expect Chinese vendors to gradually benefit from replacement demand given attractive pricing and likely favourable policies.

Who will be the winners?

We identify seven companies that should benefit most from this megatrend. We still like leading camera sensor players, which we think benefit from the widest range of ADAS applications (Tung Thih, Panasonic, Nidec & Wistron NeWeb). We expect HMI suppliers’ increasing value-add to secure steady long-term growth (Joyson, iFlytek & Baidu). We think those with expertise in AI will see increasing opportunities as the market develops (Baidu, Panasonic & iFlytek). Also, we highlight 27 Asian corporates with potential autonomous driving exposure (page 6).

BNPP recommendations

Note: Priced at close of business 29/9/2016. Share prices and TPs are in listing currency. Sources: FactSet; BNP Paribas estimates

Our research is available on Thomson One, Bloomberg, TheMarkets.com, FactSet and on http://eqresearch.bnpparibas.com/index. Please contact your salesperson for authorisation. Please see the important notice on the inside back cover.

PREPARED AND PUBLISHED BY NON-US BROKER-DEALER(S): BNP PARIBAS SECURITIES (ASIA) LTD, BNP PARIBAS SECURITIES (JAPAN) LTD. THIS MATERIAL HAS BEEN APPROVED

FOR U.S DISTRIBUTION. ANALYST CERTIFICATION AND IMPORTANT DISCLOSURES CAN BE FOUND AT APPENDIX ON PAGE 108

Jimmy Huang Vey-Sern Ling Laura Chen [email protected] [email protected] [email protected]

+852 2825 1124 +852 2825 1802 +852 2825 1280

Masahiro Wakasugi [email protected] +813 6377 2240

Company BBG code Share price Current Previous Current Previous %change Up/downside

Panasonic Corp 6752 JP 1,037.50 BUY Buy 1,400.00 1,400.00 +0.0% +34.9%

Baidu BIDU US 182.23 BUY Buy 230.00 195.00 +17.9% +26.2%

Joyson Electronics 600699 CH 33.73 BUY Buy 41.00 41.00 +0.0% +21.6%

Tung Thih Electronic 3552 TT 448.00 BUY Buy 527.00 657.00 -19.8% +17.6%

Nidec Corp 6594 JP 9,323.00 BUY Buy 10,700.00 10,700.00 +0.0% +14.8%

Wistron NeWeb 6285 TT 98.00 BUY Buy 112.00 92.23 +21.4% +14.3%

Iflytek 002230 CH 28.43 HOLD - 31.00 - - +9.0%

Rating Target price

mailto:[email protected]




ASIA TECHNOLOGY Jimmy Huang

BNP PARIBAS 30 SEPTEMBER 2016

Investment thesis

Autonomous driving is a goal for leading technology giants and automakers with the likes of Apple, Google, Mercedes Benz, BMW and Tesla aggressively investing in this area. We expect this to be a mega trend in China, driven by favourable ADAS regulation, which we expect in the near future.

Autonomous vehicles rely on advanced driving assistant systems (ADAS) and connected car systems. There are five critical component areas: 1) sensors or radars, 2) deep learning (AI), 3) high definition mapping, 4) human machine interfaces (HMI), and 5) motion control. In related Greater China supply chains, international vendors dominate, but we expect Chinese vendors to benefit from replacement demand given attractive pricing and likely favourable policy.

We expect seven companies to benefit most from this mega trend (see front page chart). We still like leading camera sensor players, which we think benefit from the widest range of ADAS applications. We expect HMI suppliers’ increasing value-add to secure steady long-term growth. We think companies with expertise in AI will have opportunities as the market develops. We also highlight 27 Asian corporates with potential autonomous driving exposure (see Exhibit 1).

Catalyst

1 Smart car models introduced by global Auto OEMs.

2 New policies launched by governments to support the development of ADAS and autonomous driving.

3 Breakthroughs in automotive technologies such as sensing algorithms and artificial intelligence.

Risk to our call

Risks to our call include 1) worse-than-expected smart car market growth, 2) a lack of government regulation support, and 3) slower-than-expected smart car development.

CONTENTS Investment summary .................................................................... 3

Stock recommendations ............................................................... 4

Autonomous driving – an automotive revolution .................... 7

ADAS – a fast growing global market ...................................... 11

Connected car system market also picking up ...................... 15

Key components in the autonomous driving space .............. 17

The opportunity for Chinese suppliers .................................... 26

Financial comparison .................................................................. 29

Wistron New Web (Laura Chen) ................................................ 31

Not-rated company visits .......................................................... 33

Appendix ........................................................................................ 37

Company reports ......................................................................... 40

Event calendar

Date Event

Oct 2016 iFlytek 3Q16 earnings release

Oct 2016 Baidu 3Q16 earnings release

Nov 2016 Tung Thih 3Q16 earnings release

Nov 2016 Wistron NeWeb 3Q16 earnings release

Oct 2016 Nidec 2Q17 earnings release

Oct 2016 Panasonic 2Q17 earnings release

Oct 2016 Joyson 3Q16 earnings release

Abbreviation Terms

ADAS Advanced Driving Assistant Systems

PA Park Assist

LDW Lane Departure Warnings

BSD Blind-Spot Detection

FCW Forward Collision Warning

HMI Human Machine Interface

ESC Electronic Stability Controls

ESP Electronic Stability Programmes

TPMS Tyre Pressure Monitoring Systems

ACC Adaptive Cruise Control

NV Night-Vision

AEB Automatic Emergency Braking

LKA Lane Keeping Assist

TJA Traffic Jam Assist

2



Investment summary

Autonomous driving is a key goal for OEMs and international tech giants

Autonomous driving vehicles appear to be quickly becoming a key goal for global OEMs and leading technology companies. In China, companies including Baidu, FAW, SAIC, BAIC, Chery, and Changan have shown an interest in developing their own models with targeted mass production in FY21 at the earliest. The benefits of autonomous driving include reducing accidents and improving conditions for the likes of senior and disabled drivers. Although we believe China is behind other countries in terms of government regulation and technology know-how, we expect the Chinese government to announce ADAS-related policy by mid-FY17 and for this to boost local demand. This report focuses on the Chinese situation and key Asian stock ideas.

Autonomous driving = ADAS + connected car systems

Autonomous driving means vehicles leveraging various equipment and systems to replace the functions of the human brain, eyes, ears and hands. As such, autonomous vehicles (AV) rely heavily on two systems: 1) advanced driving assistant systems (ADAS) and 2) connected car systems. ADAS is designed to avoid dangers and improve the driving experience. Strategy Analytics forecasts the global ADAS market (in shipments) to see a 20%+ CAGR over 2014-19. ADAS comprises a number of systems, of which we think park assist (PA) is most common, then lane departure warnings (LDW), blind-spot detection (BSD) and forward collision warning (FCW). Connected car systems serve functions like in-vehicle services (navigation, smart-device access, entertainment, etc), vehicle management (remote access to locks, horns, etc) and, most importantly, vehicle-to-vehicle (V2V) communication.

Five key technologies that should benefit

Autonomous driving vehicles comprise a wide range of auto components and communication systems. We highlight five areas that we expect to play critical roles in technology development: 1) sensors and radars, 2) deep learning (AI), 3) high definition mapping, 4) human machine interfaces (HMI), and 5) motion control. Generally, ADAS covers the three major processes of ‘seeing’, ‘thinking’ and ‘taking action’ with various types of sensor/radar taking in information from the environment, deep learning (AI) processing the data through complex algorithms enabling modules to make decisions, which are executed through motion control. Meanwhile, both HMI and high definition mapping help users interact with control systems and help decision-making based on a vehicle’s surroundings.

A long-term opportunity for China’s local supply chain

We think our research is differentiated by our identifying suppliers in Greater China – through company visits and supply chain checks – that we expect to act more like component suppliers than system providers like many international players. China has a relatively low penetration of auto electronics, including electronic stability controls/programmes (ESC/ESP), electronic brake assist, tyre pressure monitoring systems (TPMS) and even navigation (source: Strategy Analytics). However, we expect demand in China for ADAS and connected car systems to grow in the next five years driven by regulation and rising safety concerns. We expect this to trigger demand for components. Longer term, we expect replacement demand to pick up in China and expect companies providing quality products to benefit most. Currently, Chinese companies have very limited AV sales exposure, so it is not easy to find pure-play AV stock ideas. Instead, we have identified companies with niche advantages and leading positions in various spaces.

3



Stock recommendations

Within Greater China’s autonomous driving vehicle supply chain we select by market position and future opportunity, taking into account valuations and growth potential.

Joyson Electronics – reiterate BUY with RMB41 TP (Jimmy Huang)

We reiterate our BUY rating with a P/E-based TP of RMB41. We like Joyson’s leading position in auto electronics, particularly HMI, and expect it to benefit from ongoing safety component demand. Joyson entered the ADAS market with its acquisition of KSS in 1H16, joining its existing customers including Volvo, Hyundai and Renault, and strengthening its market position to capture HMI future trends like large-size touch panels, multifunctional controllers and heads-up displays by its acquisition of TS (1H16). Joyson remains positive about receiving approvals from authorities for a potential fund raising in 4Q16, which would enable it to ease its financial expense burden. Downside risks to our view: 1) soft global auto market demand, 2) unfavourable currency trends, especially EUR depreciation, 3) intensified market competition, and 4) difficulties integrating acquired overseas companies.

Baidu – reiterate BUY with USD230 TP (Vey-Sern Ling)

We think Baidu is well positioned in three of the five key components required for autonomous driving: 1) deep learning (AI); 2) high definition mapping; and 3) HMI. Leveraging its foundational work on AI, Baidu’s autonomous driving system was successfully tested on Beijing’s public roads in Dec 2015, and is now undergoing further road tests in Wuhu, Shanghai, Wuzhen and California. Baidu’s navigational map is currently the most used mobile map in China, and its co-investment with Ford in Velodyne provides its self-developed 3D high-precision map with centimetre-level positional accuracy. On HMI, Baidu (with its voice recognition technology) is working with major automobile OEMs to provide a suite of telematics products such as voice-based navigation and digital assistants in cars. We reiterate our BUY rating on Baidu and raise our TP by 18% to USD230 to reflect to reflect revenues from potential AI commercialisation opportunities from 2018.

Wistron NeWeb – reiterate BUY with TWD112 TP (Laura Chen)

We reiterate our BUY rating and lift our earnings forecasts by 6.3% for 2017 and 10.2% for 2018 to reflect solid shipments in smart-home-related and auto-related business. We believe WNC’s growth outlook is intact both near and longer term on rising demand for connected cars, M2M and smart-home applications. We raise our TP to TWD112 (from TWD92.23) based on 14x our 2017E EPS (multiple unchanged, but rolled over to our 2017 estimate). For 4Q16, with a new project ramping up in its smart-meter-related business in Europe and continued momentum in smart-home and auto-related business, we forecast revenue of TWD15.7b (+11.5% q-q; +9.6% y-y; vs BBG consensus of TWD15.4b) with a steady GM of 13.3% and OPM of 4.6% leading to net profit of TWD613m (+22.0% q-q; +20.2% y-y).

Tung Thih – reiterate BUY with TWD527 TP (Laura Chen)

We reiterate our BUY rating with a DCF-based TP of TWD527. We expect TTE to benefit from increasing content per car from global ADAS and TPMS demand driven by rising awareness of auto safety in developed countries leading to more mandatory products for new vehicles. Although Chinese authorities have not yet set compulsory standards for ADAS and TPMS for new vehicles, we expect China will soon follow the US and Europe on such standards to improve auto safety. We believe TTE will gain from incremental Chinese demand given its solid Chinese customer base, varied product offerings, competitive cost structure and localization benefits. Also, we expect GM’s global D2XX platform to serve as secular growth driver.

iFlytek – initiating at HOLD with RMB31 TP (Jimmy Huang)

We initiate at HOLD with a target price of RMB31, implying 9.0% upside. Founded by doctors from the University Science and Technology of China, iFlytek is a leading voice recognition expert in China with main applications in automotive and education. We expect it to benefit from developments in autonomous driving and artificial intelligence. iFlytek posted 42%/33% revenue/earnings CAGRs in FY10-15; we forecast 40%/34% earnings CAGRs over the next three years (FY16-18E). We see

4



iFlytek as a beneficiary of AV trends, but are concerned about competition from Baidu and Nuance, and a rich valuation (52x FY17E P/E) – hence our HOLD rating.

Nidec – reiterate BUY with JPY10,700 TP (Masahiro Wakasugi)

Nidec’s main products relate to sensors/radars and motion control. It acquired Nidec Elesys in 2014, a strong total-ADAS-solution provider supplying not only cameras and radars for sensing but also ECUs. We believe Nidec will benefit as the market moves towards autonomous driving. Also, we expect Nidec’s motors and ECUs to play a key role in motion control. Nidec is a leading supplier of electric power steering (EPS) motors, and it intends to expand motors for electric brakes (EB) for global tier-one makers. We believe that Nidec’s ADAS products and high-efficient motors with ECUs will lead to strong earnings growth in the foreseeable future.

Panasonic – reiterate BUY with JPY1,400 TP (Masahiro Wakasugi)

Panasonic has several competitive products for sensors/radars, deep learning (AI), and HMI. It is particularly strong in HMI in areas like infotainment and e-cockpit systems. The company has already been a major supplier for rear-view cameras and ultra-sonic sensor system for ADAS. Panasonic has been developing autonomous parking systems with deep learning technology (as we discussed in our report Can deep learning save Japanese electronics? 7 Apr 2016). Its strength is in providing integrated products which combine camera/sensors, infotainment systems and deep learning technology. We believe Panasonic will be a key solution provider for ADAS and autonomous driving systems. According to the Nikkei and several other press reports, Panasonic will jointly produce core EV components with Beijing Automotive Group (BAIC) and will jointly build a factory in Tianjin. We think this cooperation may expand to other areas, including ADAS and related products. We think it’s worth closely monitoring how well Panasonic can grow its automotive business in China.

5

https://eqresearch.bnpparibas.com/researchfs/FS__7919ef170a304af18cdb5ffcfee547c2.pdf

https://eqresearch.bnpparibas.com/researchfs/FS__7919ef170a304af18cdb5ffcfee547c2.pdf



Exhibit 1: Companies in Asia with exposure to ADAS/autonomous driving theme

Name Chinese BBG code Mkt cap Daily trading volume (6M) Product/area (autonomous driving)

(USD m) (USD m)

Greater China

Baidu* 百度 BIDU US 63,097 550 AI for autonomous driving

Sunny Optical* 舜宇光学 2382 HK 5,382 27 Camera lenses for ADAS sensing functions

Minth 敏实集团 425 HK 3,929 9 Cameras for ADAS sensing functions

Wistron Neweb* 啟碁科技 6285 TT 1,071 8 Communication modules for autonomous driving

Tung Thih* 同致電子 3552 TT 1,203 24 TPMS sensors and CMOS cameras for ADAS

Tong Yang 東陽實業 1319 TT 1,393 6 Collision parts for ADAS functions

Cub Electparts 為升電裝 2231 TT 914 5 TPMS sensors for ADAS functions

Hikvision* 海康威视 002415 CH 22,355 57 Cameras for ADAS sensing functions

Joyson* 均胜电子 600699 CH 3,470 84 Auto HMI solutions for ADAS/autonomous driving

iFlytek* 科大讯飞 002230 CH 5,513 115 Voice control HMI for ADAS/autonomous driving

O-Film 欧菲光 002456 CH 5,736 117 Cameras for ADAS sensing functions

Jingu 金固股份 002488 CH 1,283 96 Camera sensors for ADAS sensing functions

Tatfook Tech 大富科技 300134 CH 3,056 119 Vision sensor solutions for ADAS sensing functions

Navinfo 四維圖新 002405 CH 3,555 141 Electronic mapping for autonomous driving navigation

Hi-target Navigation 中海達 300177 CH 943 27 Autonomous driving navigation

APG 亞太股份 002284 CH 1,812 124 ADAS functions/components: ABS, AEB and ACC

Vie Science & Technology 万安科技 002590 CH 1,765 71 ADAS functions/components: ABS and EBS

Dongfeng Electronic Tech 东风电子 600081 CH 708 18 ADAS functions/components: ABS

Toupu 拓普集团 601689 CH 2,724 35 Intelligent braking systems for ADAS functions

Wanxiang Qianchao 万向钱潮 000559 CH 4,988 75 Vehicle control systems for ADAS functions

Neusoft 東軟集團 600718 CH 3,207 84 In-vehicle infotainment for ADAS/autonomous driving

Japan & Korea

Panasonic* 6752 JP 24,281 91 ADAS sensors and deep learning for autonomous driving

Nidec* 6954 JP 34,190 162 Camera sensors/ radars for ADAS sensing functions

Denso 6902 JP 31,164 82 ADAS sensors

LGI* 011070 KS 1,672 14 Camera modules/sensors for ADAS sensing functions

Mcnex 097520 KS 182 5 Camera modules for ADAS sensing functions

Mando 204320 KS 2,373 14 ADAS functions/components: ABC, ESC and ACC

*BNP Paribas coverage; Mkt cap and trading volume are as of 29 Sep 2016 Sources: Companies; Bloomberg

6



Autonomous driving – an automotive revolution

Autonomous driving appears to be a key goal for global OEMs and their customers, and major auto players and technology companies have already demonstrated autonomous driving to the public. SAE International, a US-based association focusing on transport industries, defines five levels of automation for vehicles with the performance of longitudinal and lateral tasks with limited to no driver control.

We note that autonomous driving relies heavily on two systems: 1) advanced driving assistant systems (ADAS) and 2) connected car systems. While China lags other countries in this space, we expect it to catch up quickly with long-term demand driven by government policies that we expect to be implemented in the near future.

What’s so good about autonomous driving?

If full autonomous vehicles are achieved, we summarize four key advantages:

1 Reduced accidents: Autonomous vehicles can reduce human errors in driving by avoiding human carelessness, distraction and misjudgement.

2 Driver comfort: Drivers can relax more with partially autonomous driving and, in the future, hand over full driving operation to the vehicle.

3 Ease for senior and disabled drivers: Autonomous vehicles enable elderly and disabled drivers to have private and convenient transportation.

4 Less traffic, pollution, etc: Autonomous vehicles can help ease traffic congestion through good communication between vehicles. This helps minimize fuel consumption and emissions, as does efficient driving.

Players in the US, Europe and Japan are ahead in autonomous driving area

Autonomous driving is a revolution for the automotive industry. Globally, major auto OEMs including BMW, Audi, Mercedes Benz, Volvo and Tesla have started to develop technology for autonomous driving with most targeting launches in 2020 (Exhibit 3). Furthermore, innovation for autonomous driving has changed the competitive landscape with many technology giants such as Google, Apple and Baidu entering this area given the requirement for more advanced technologies like sensor monitoring and algorithm computing.

Exhibit 2: Level of automation of vehicle

Level Degree of automation Driver Automation system Example

0 Driver only Driver continuously in control of speed and direction. No intervening vehicle system active N/A

1 Assisted Driver continuously performs the longitudinal or lateral dynamic driving task.

The other driving task is performed by the system. Park assist

2 Partial automation Driver must monitor the dynamic driving task and the driving environment at all times.

System performs longitudinal and lateral driving task in a defined use case.

Traffic jam assist

3 Conditional automation

Driver does not need to monitor the dynamic driving task nor the driving environment at all times; but must always be in a position to resume control.

System performs longitudinal and lateral driving task in a defined use case. Recognises its performance limits and requests driver to resume the dynamic driving task with sufficient time margin.

Highway patrol

4 High automation Driver is not required during defined use case. System performs the lateral and longitudinal dynamic driving task in all situations in a defined use case.

Urban automated driving

5 Full automation No driver required. System performs the lateral and longitudinal dynamic driving task in all situations encountered during the entire journey.

Full end-to-end journey

Sources: SAE; BNP Paribas

7



China – the current situation: major players and regulations

In China, we think key players are Baidu, FAW, SAIC, BAIC, Chery, Changan and some major local OEMs that have set up alliances with R&D institutions to develop related technology. Baidu has launched its autonomous driving project, targeting mass production by 2021. We expect ongoing demand growth in China, led by ADAS demand, development of major OEMs and government regulation.

In contrast to leading Western countries, the Chinese government has not set up any regulations around safety systems, although local supply chain companies expect this to happen in 2017 at the earliest.

There are also major OEMs who have developed alliances with various research institutions to pursue autonomous driving technology (Exhibit 5).

Exhibit 3: International players’ plans on autonomous driving

Name Development Cooperation partner Launch timetable

Google Accomplished 1.5m miles autonomous driving without accidents Chrysler Planned to launch in 2020

Apple Speculation around launch of project "Titan" for autonomous driving (as reported here)

n/a n/a

Tesla Autopilot system allows ADAS functions including ACC, AEB, auto parking, and highway autonomous driving.

n/a Launched in 2017

Volvo Autonomous driving testing in Beijing on Mar 2015 with a driving distance of 1,200km

Microsoft, Ericsson Commercialized in 2020

BMW Developing Project i 2.0 for autonomous driving vehicle "iNext" series Baidu iNext expected to be launched in 2020

Mercedes Benz Obtained autonomous driving licence for passenger vehicles in Nevada, US; and finished highway autonomous driving in Germany

n/a n/a

Audi Demonstrated autonomous driving concept n/a Expected in 2018

Sources: Companies; BNP Paribas

Exhibit 4: Key Chinese domestic players’ plans on autonomous driving Name Development Cooperation partner Launch timetable

Baidu Baidu launched its first autonomous driving on-the-road testing in Beijing, China in Dec 2015. It has also set up a government-approved operation area for autonomous driving vehicles in Wuhu, China

Chery, BMW Commercialized in 2018; Mass production in 2021

FAW Accomplished highway autonomous driving test in Changsha National University of Defense Technology, China Mobile, Datang Telecom

n/a

SAIC Demonstrated its 2nd generation smart car in 2015 Avi China, Alibaba, Navinfo Target to equip highway autonomous driving function on SAIC vehicle by 2020

BAIC Research on smart car since 2013 Beijing Union University, Leshi n/a

Chery Developed smart car with Wuhan University since 2013; signed contract with Asia-Pacific Mechanical on smart car components supply

Wuhan University, Asia-Pacific Mechanical

n/a

Changan Demonstrated autonomous driving vehicle sample in 2015 Huawei, Tsinghua University Target to start mass production by 2030


Exhibit 5: OEM alliances with internet/electronic companies for autonomous technology

R&D stage Technology Testing stage On-road test & trial stage

Changan Chery – Wuhan University FAW – The National University of Defence Technology (2011)

Geely BAIC – LeTV Beijing Hyundai – The Military Transportation University (2012)

Dongfeng – Huawei SAIC – Alibaba Baidu – BMW (2015)

GAC – Hefei Institute BYD – Beijing Institute of Tech

Source: BNP Paribas

8

http://www.macrumors.com/roundup/apple-car/



Government regulations helping to develop autonomous driving technology

Government policy and regulations in various countries have helped the development of autonomous vehicle technology. As summarised below, ADAS products are starting to become mandatory in new vehicles in some markets like the US, EU and Japan. While China is lagging behind, we expect regulations to be implemented in the near future. Similarly, China’s NCAP (New Car Assessment Programme) plans to adopt active safety technologies in its rating system, and this could also support China’s ADAS market.

US: At the state level, some authorities have legalised autonomous driving for various applications and conditions. At the federal level, the National Highway Traffic Safety Administration (NHTSA) has issued relevant policies and has set timelines for compulsory adoption of ADAS systems (including FCW, LDWS, RVS CIBS and DBS) in new passenger vehicles.

European Union: EU governments have set timetables for compulsory ADAS systems (including ESC, AEB and LDWS) in new passenger and commercial vehicles.

Japan: The Japanese government started giving special permission to OEMs in relation to autonomous driving on public roads in FY13.

China: The Chinese government has not yet passed any requirements or policies on autonomous driving vehicle but there are expectations in the market for this to happen in 2H16 or 1H17. According to local news, China’s MIIT is planning to launch its “Technology Roadmap for Internet of Vehicles” in the near term. This is likely to trigger further demand for smart cars in China, as smart cars play an important role in the “Internet of Vehicles” space.

ADAS and connected car systems are critical to autonomous driving

An autonomous vehicle can be seen as a form of transport leveraging equipment and systems to replace the functions of the brain, eyes, ears and hands. We think autonomous driving technology relies most heavily on two systems: 1) advanced driving assistant systems (ADAS) and 2) connected car systems.

Exhibit 6: Global development of active safety regulations and ratings

Sources: Continental; BNP Paribas

9

http://nev.ofweek.com/2016-09/ART-71008-8120-30041386.html

http://news.xinhuanet.com/auto/2016-07/20/c_129161424.htm



The challenge for autonomous vehicles today

There are plenty of challenges for autonomous driving. The major issues, in our view, are the lack of legislation and infrastructure, and standardization of hardware and software. We think key challenges for autonomous driving in China include:

1. The absence of specific regulations. China has not yet established specific autonomous vehicle regulations, although the market expects MIIT to release related policy in 2H16.

2. Inadequate infrastructure and maps. Autonomous driving needs infrastructure such as intelligent road systems, road signs and precise, real-time maps to support operations. Although we think China’s roads are being mapped, we think the process might take at least another three to five years.

3. Unpredictable driver behaviour. In China, we note a lot more unpredictable driving than in other developed markets, which significantly increases the difficulties for systems handling or responding to those uncertainties.

4. Bad road conditions. For autonomous driving to operate well, there needs to be well-established highways with satisfactory markings. Road conditions in China, particularly in rural areas, are not up to the standards of more developed markets. We think poor road conditions increase the risks of autonomous driving.

5. Others challenges. We think other challenges include: 1) Security, as communications can be manipulated by hackers or terrorists, 2) the need for expensive testing facilities, and 3) the lack of global standards given countries have varying road traffic laws and different levels of technology advancement.

Exhibit 7: Autonomous driving versus human driving

Sources: BNP Paribas

Exhibit 8: Autonomous driving mainly supported by ADAS and connected car

Sources: BNP Paribas

10



ADAS – a fast growing global market

Through the years, auto OEMs have improved driving experience and driver safety by embedding innovative technologies into vehicles. To start with, simple passive safety products, such as airbags and seatbelts, helped alleviate damage from accidents. As technologies moved from mechanical to electronics, more innovative intelligent products/systems became available, including advanced driver assistance systems (ADAS), which helps prevent crashes as well as minimize their impact.

ADAS explained in brief

ADAS are advanced and intelligent automotive systems designed to avoid potential dangers and improve the driving experience. We see six major functions in ADAS: adaptive cruise control (ACC), lane departure warnings (LDW), blind-spot detection (BSD), forward collision warning (FCW), night vision (NV) and park assist (PA) (see the appendix for more detailed definitions).

ADAS can be divided into three processes: 1) seeing – sensing the environment using cameras, radars and ultrasonic sensors; 2) thinking – making decisions on dynamic data from sensors or mapping systems; and 3) action – motion control (like steering, braking, accelerating and suspension setting) under various conditions.

ADAS currently limited to higher-end models due to costs

ADAS is currently limited to higher-end models simply because of costs. The safety systems, especially the more advanced ones within the ADAS field, can easily add more than USD2,000 per vehicle. The ASP of ADAS products is likely to fall less than 5% annually (except for the function of night view), based on Strategy Analytics data, which is less than is seen in most consumer electronic components.

Exhibit 9: Three steps in ADAS

Source: BNP Paribas

Exhibit 10: Safety system ASP

Source: Strategy Analytics

See/Perception ActionThink

Sensor

LidarCameraRadar

UltrasoundInfrared

Electronic Control Unit (ECU)

Processor

Human Machine Interface

Head-up displayAudio

Control systemEngine

SteeringBrake

0

100

200

300

400

500

2013 2014 2015 2016E 2017E 2018E 2019E 2020E 2021E 2022E

(USD) Parking (With Camera) Parking (Surround View)Blindspot LDWSDistance Warning Night viewTyre pressure warning

11



A fast growing global market; with parking assist the most common function

With rising concerns over safety and technology development, ADAS has been a fast growing area; Transparency Market Research expects a 19.2% CAGR in ADAS volumes in 2014-20. Of ADAS functions, park assist (PA) is the most common, based on Frost & Sullivan’s data, with a 62% ADAS shipment volume market share in 2012. We see the strongest growth in LDW, BSD and FCW driven by regulations and their benefits once they become well understood.

Exhibit 11: Global ADAS market value Exhibit 12: System demand growth CAGR 2014-19

Source: Strategy Analytics Source: Strategy Analytics

Exhibit 13: ADAS market outlook by shipment Exhibit 14: ADAS market share outlook by shipment

Sources: Frost & Sullivan; BNP Paribas Sources: Frost & Sullivan; BNP Paribas

Exhibit 15: Safety system demand – 2015 vs 2020 (shipments)

Sources: Strategy Analytics; BNP Paribas

0

5

10

15

20

25

30

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

45,000

2015

2016

E

2017

E

2018

E

2019

E

2020

E

2021

E

2022

E

2023

E

(y-y %)(USD m) Market Size (LHS) Growth (RHS)

0

5

10

15

20

25

30

Con

v. E

ngin

e

HE

V/EV

Tran

smis

sion

Cha

sis

Cov

. Saf

ety

ADAS

Body

Secu

rity

Driv

ier i

nfo

Tota

l

(%)

0

5

10

15

20

25

ACC LDW BSD FCW NV PA

(m units)2012 2020E

10.8 17.19.5

22.611.5

17

6.5

22.6

5.2

61.7

15.5

0

10

20

30

40

50

60

70

80

90

100

2012 2020E

(%) ACC LDW BSD FCW NV PA

020406080

100120140160180

Fron

t

Side

impa

ct(F

ront

)

Side

impa

ct(R

ear) Oth

er

Occ

upan

tde

tect

ion

Rev

. Bel

t Ten

s.

Col

lisio

nw

arni

ng

TPM

S

E-ca

ll te

lem

atic

s

Dro

wsi

ness

mon

itor

Nig

ht V

isio

n

Oth

er A

DAS

(m units) 2015 2020E

Air bag

12



Chinese ADAS market – significant potential

Although China is one of the biggest automotive markets in the world, its ADAS penetration rate is only in the low single digit percentages, according to Sohu data. According to Frost & Sullivan, the China ADAS market value was USD430m in 2013, around just 20% of North America’s. We see a significant growth opportunity in China especially given the low base. Frost & Sullivan forecasts the China ADAS market (in dollar value) will grow at a 72% CAGR from 2013 to USD2.2b in 2020.

We think China’s ADAS market is currently dominated by international OEMs and component suppliers, based on our supply chain check. The international OEMs include BMW, Daimler, GM, Toyota and Volkswagen, and represent 72% of China’s ADAS market with key components coming mainly from Bosch, Continental, TRW, Autoliv, Denso and Delphi, according to Frost & Sullivan.

We expect China ADAS demand to pick up significantly, which will also result in more opportunity for local ADAS components suppliers, on the back of lower pricing with decent quality, in our view. Currently, we see majority of China domestic players participate more in the single component manufacturing, instead of a whole system.

Exhibit 16: History of driving assistance systems

Sources: BCG; BNP Paribas

Exhibit 17: Global ADAS component market share (2010) Exhibit 18: China ADAS market size and y-y growth

Source: Continental Source: IHS

Cruise control1958

Seatbelt reminder

1970s

Antilock braking system1971

Airbags1973

Electronic stability control1987

Forward collision

warning/assist2000/2008

Night vision2000

Rear camera/Park

assist2002

Lane departure

warning/lane keep assist2005/2014

Adaptive front lights/Automat

ic parking/ Blind spot detection

2006

Surround view systems

2007

Drowsiness alert2010

Single lane highway

2016

Autonomous valet

parking/Traffic jam autopilot

2017

Highway autopilot with lane changing

2018

Urban autopilot

2022

Safety and convenience(1950-2000)

ADAS(2000-present)

Partially autonomous(2016-2025)

Autonomous(2025~)

Continental20%

Denso18%

Bosch10%Autoliv

7%

Delphi7%

Gentax6%

Valeo6%

Hitachi6%

Hella5%

Honda Elesys

4%

Others11%

0

5

10

15

20

25

30

35

0

2

4

6

8

10

12

14

16

18

20

2012 2013 2014 2015E 2016E 2017E 2018E 2019E

(y-y %)(RMB b) Market Size (LHS) Growth (RHS)

13



Exhibit 19: Global Auto OEMs and components supplier for ADAS function

OEM ---------------- Current Available ---------------- -------------------------------------------- Future Offering -------------------------------------------- Supplier Tie-up

ADAS function SAE level ADAS function SAE level Year Model

Audi Traffic Jam Assist Level 2 Traffic Assist,

Piloted Driving, Piloted Parking

Level 3 2017 onwards A8, A7, Q8 Valeo, Mobileye, Continental, Bosch

BMW Traffic Jam Assist, Assistive Parking Level 2

Active Assist, Remote Valet

Parking Level 3 2018 onwards 7-series and 5-

series Continental, ZF Lenksysteme

Cadillac AEB, ACC Level 1 SuperCruise Level 3 2020 onwards CTS and Escalade

Mostly in-house, TRW, Laird Tech

Ford ACC, AEB Level 1 Active City Stop Level 2 2017 onwards Fusion and Escape

Continental, Velodyne, Bosch, In-house

Mercedes-Benz Traffic Jam Assist Level 2 Distronic Plus with Steer Assist Level 3 2019 onwards S-Class and E-

Class Quanergy (For R&D)

Tesla Autopilot Level 2 Autopilot 2.0 Level 3 2019 onwards Model S, X3 Mobileye, Bosch, nVidia

Volvo Pilot Assist Level 2 City Safety Level 3 2022 onwards XC90, S90 Continental, Autoliv

Source: Frost Sullivan; Company data, BNP Paribas

14



Connected car system market also picking up

A connected car connects devices to other devices in the vehicle or to devices, networks and services outside the vehicle (eg, other vehicles, the driver’s home or office network or infrastructure network).

Connected car systems provide functions such as in-vehicle content and services (eg, navigation, smart device access and entertainment), vehicle management (remote access to locks and horns), V2V communication and even autonomous driving. Major components and technologies are HMI, machine learning algorithms, internet connectivity and telematics service.

Frost & Sullivan forecasts a 21% CAGR over 2015-22 in connected car units from 18.8m in 2015 to 69.4m in 2022 driven by mobility services (eg, parking, route planning and navigation).

While connected car systems include various functions (see appendix for details), we think a good connected car system requires advanced hardware and machine learning algorithms. We expect future trends to include:

1 High speed connectivity: We believe high speed connectivity technologies, such as LTE or 5G, are very important to connected vehicle systems given increasing demand from navigation and high-speed downloads.

2 Smartphone interfaces: Vehicles are increasingly adopting smartphone interfacing technology, including Android Auto, CarPlay and MirrorLink.

Exhibit 20: Global connected car unit forecasts, 2015-22

Source: Frost & Sullivan

Exhibit 21: Smartphone interface technology penetration (2015 and 2020)

Sources: SA

0

10

20

30

40

50

60

70

80

2015 2016E 2017E 2018E 2019E 2020E 2021E 2022E

(m units) North America Europe China Brazil South Korea Japan India

0

10

20

30

40

50

60

70

80

90

MirrorLink Android Auto CarPlay

(%)2015 2020

15



3 Navigation. We expect a standardized digital mapping system which can enable up-to-date local information in order to seek the most efficient route.

4 AR/VR. More and more auto OEMs are using augmented reality (AR) and gesture technology in the HMI area.

Chinese connected car features are far behind other countries’

Connected car development in China is far behind that in other regions (particularly the US and Europe) with low or no penetration in various features, notably connected infotainment and telematics. We attribute this to average lower specs in the Chinese auto market, cost concerns and a lack of strong promotion from major OEMs.

Exhibit 22: Connected car features

Features Europe North America China

Connected Infotainment Feature Door to Door Navigation M L NA

Open Sourcing of Apps (SDK) L H NA

Apple CarPlay M M L

Android Auto M M L

MirrorLink M L L

Connectivity Standards (USB, BT, Wi-Fi, NFC) L M L

Wireless Device Charging NA L NA

Wireless Hotspot L M NA

Embedded LTE L M M

App Stores L M L

Navigation Map Over the Air Updates (NDS) L M NA

V2V/V2X L M L

Mobility Services (parking, Route planning, Nav) H L NA

Telematics Feature Telematics Safety: Automatic Collision Notification (ACN), eCall, Roadside Assistance L H L

HWW and Conditional Monitoring L H L

Smart and Personalized Access with Biometrics M M L

Telematics-based Insurance: Mileage-based, pay-how-you-drive (PHYD)-based Insurance H M L

Prognostics Services L M NA

Recall Management L M NA

Software Over-the-Air Updates L M NA

Firmware Over-the-Air Updates L L NA

Personalized Concierge Services L M NA

Impact Alert (for emergency and insurance) L M NA

VRM and CRM L M L

Geo-Fence Alerts and Marketing L L NA

Remote Function App with Parking L M L

Input and Output HMI Feature Advanced Voice Recognition M M M

Natural Speech Recognition M M M

Capacitive Touch Screen L M L

Touch Gestures L M NA

Multifunctional Controller M M L

Touchpad with Handwriting Recognition M L M

Hand/Air Gestures NA NA NA

12" Plus Central Display L M L

Head-up Display – Combiner L L NA

Voice personal Assistants L M M

Haptic Touch Feedback NA L NA

Portrait Central Screen L L NA

* Impact: NA = Not Available; L = Low; M = Medium; H = High Source: SA

16



Key components in the autonomous driving space

Autonomous driving is comprised of a wide range of auto components and communication systems. We categorize components into ADAS and connected car areas, and discuss the potential opportunity in China. There are five critical component areas: 1) sensors or radars, 2) deep learning (AI), 3) high definition mapping, 4) human machine interfaces (HMI), and 5) motion control.

Other components also matter

As well as these five key component areas, other components are also important, for example a 48V power system, various chipsets and 5G connectivity.

48V architecture seems ideal given the increasing amount of powertrain and chassis electrification needed for autonomous driving. Currently, European OEMs, including VW, Audi, Bentley, Citroen and Peugeot, are leading this trend, while we expect the US and Asia to follow suit.

IC chipsets support both ADAS and connected car systems. Automotive chip suppliers NVIDIA, Renesas, NXP, Texas Instruments, STMicroelectronics (silicon supplier to Mobileye), QCOM and Infineon have all shown their commitment to auto electronics. Furthermore, we believe the performance of neural-network-based AI systems is directly correlated with network scale and the volume of useful data fed into them, all requiring heavy GPU usage.

5G is also vital in autonomous driving as vehicles require real-time sensors, different services, non-stop machine learnings processes, and integration with different apps and devices (V2V and V2I). Compared with 3G or 4G LTE, 5G systems can provide not only basic phone calls, messaging, data transfers and cloud computing, but also advanced video streaming, voice control, high speed connectivity and IoT services.

However, we do not focus on these spaces given there are few players with related exposure in Greater China. Next we provide overviews of our five focus areas and related players.

Exhibit 23: Key components for autonomous driving vehicles

Source: BNP Paribas

17



1. Sensors and radars

Sensors are a key component as ADAS cannot function without gathering information from sensors on the vehicle. We expect ADAS demand to trigger sensor demand simultaneously. Of different types of sensor, camera-based sensors have emerged as most important with various applications already (see appendix for different sensor in ADAS functions). Our regional technology team has conducted an extensive study in the ADAS sensor area (see Deep dive shows ADAS potential by Masahiro Wakasugi and Automatic for the people by Laura Chen)

Camera sensors have a wide range of applications, such as LDW, traffic sign recognition, surrounding view and parking assist. According to Strategy Analytics, camera sensor demand will grow at a 35% CAGR in 2012-17 to 68m units in 2017. Besides camera sensors, short- and medium-range radar (SRR/MRR) is also a popular technology used by ADAS functions such as cross traffic warning, blind spot detection and rear collision warnings.

Exhibit 24: ADAS sensor demand Exhibit 25: ADAS market revenue breakdown by application

Sources: Strategy Analytics, BNP Paribas Sources: IHS, Freescale

Exhibit 26: Sensor requirements of cameras in ADAS applications

--------- Camera for viewing --------- --------- Camera for sensing --------- ---- Infrared ---- -------- Radar -------

Ultrasonic Front Rear Side Inside Front Rear Side Inside Front Inside Long

Range Short/

Midrange

Adaptive Cruise Control (ACC) ● ● ● Advanced Emergency Breaking System ◎ ● ● ◎

Parking assistance (PA) ● ● ● ◎ ● ◎ ◎ ◎

Auto High Beam ● Blind-Spot Detection (BSD) ● ● ◎ ● ● ◎ ◎

Drive Recorder ● ◎ ◎ ● Forward-Collision Warning (FCW) ◎ ● ● ● In-car Drowsiness Detection ● ● Intelligent Speed Assistance ● ◎ ● Lane-Keeping Assistance ● ◎ Night-View System (NV) ● ◎ ● ● Pedestrian Collision Warning ● ● ● ● Rear-Collision Warning ● ● Traffic Sign Recognition ● ●

*● = necessary, ◎ = sometimes present Source: Gartner

0

10

20

30

40

50

60

70

80

Bolometer Camera Infrared LIDAR LRR SRR/MRR

(m units) 2012 2017E

0

200

400

600

800

2015E 2016E 2017E 2018E 2019E 2020E

(USD m)

Driver MonitorBlind SpotRear CameraSurround ViewData FusionTraffic Jam AssistFront Camera + Ped DetectionFront Camera

18

https://eqresearch.bnpparibas.com/researchfs/CR__ddd6e7f987364db7a8ecc1dccc8b0a76.pdf

https://eqresearch.bnpparibas.com/researchfs/SRF__7bc5c54fa0c549028fac82dd25925e78.pdf



More Greater China players in camera sensor space than other types of sensors

A lot of Greater China companies have exposure to the camera sensor supply chain from single components (like lens) to modules or whole systems. We think this is mainly due to the greater opportunity. While very few are developing non-camera type sensors, the technology is not yet mature. Once it is more mature, more companies might be tempted into the space.

2. Deep learning (AI)

Deep learning in ADAS and connected car systems is considered a must-have. As discussed, ADAS generally includes three processes (seeing, thinking and action). We see “thinking” as deep learning since the system or module acquires data and processes it with complex algorithms, leading to decision making in terms of object detection, classification, segmentation, map localization, path planning, etc.

AI development is also needed to support voice control and recognition (input and output). Major AI applications in autonomous driving are: detecting and recognising multiple objects; semantic analysis of areas surrounding the vehicle; recognising and predicting actions; improving perception while reducing power consumption, vs traditional algorithms; and automatic updates and training.

Deep learning is available in a few models today, in both ADAS and infotainment (HMI). For example, in HMI, most voice recognition relies on algorithms based on neural networks running in the cloud. BMW’s 2015 7 Series uses a hybrid approach, with embedded hardware to perform voice recognition when wireless connectivity is

Exhibit 27: Comparison between different sensors

Type Detection range Cost Pros Cons Supplier

Milliwave radar 100-250m Higher 1. Not affected by weather condition 2. High penetrating

1. Technology barrier high 2. Recognition ability low

Traditional auto parts supplier: Bosch, Autoliv, Continental, TRW

Ultrasound radar

<6m Low 1. Low price 2. Matured product

Low detection range

Multiple suppliers

Infrared sensor 100-200m Higher Available on NV system

High price Autoliv, Protruly

Laser radar 100m Highest 1. 3D imaging 2. High measurement accuracy

1. Extreme high price 2. Not available for all weather condition

Velodyne, Ibeo

Camera 100m Low 1. Low price 2. Mature product 3. Strong recognition on object shape

Surrounding view is highly depending on environment

Multiple suppliers


Exhibit 28: Auto component suppliers’ capacity

--------------------------------------- ADAS application --------------------------------------- --------------------------------------- Sensor suit ---------------------------------------

Adaptive Cruise Control (ACC)

Lane Keeping Assist (LKA)

Traffic Jam Assist (TJA) Park Assist Camera

(Mono) Camera (Stereo) Radar LIDAR

Autoliv ● △ △ ◎ ● ● ● △

Bosch ● ● ● ● ● ● ● ●

Continental ● ● ● ● ● ● ● ●

Denso ● ● △ △ △ ● ● △

Delphi ● ◎ △ △ ◎ (HD CMOS) ● ●

Hella △ △ △ △ △ △ ● △

Magna △ ● △ ● △ △ △ △

Mobileye △ △ △ △ ● △ ● △

Valeo ● ● △ ● ● △ △ ●

ZF-TRW ● ● △ △ ● △ ● △

△: Limited capability; ◎: Partial capability; ●: Full capability Source: Frost & Sullivan

19



absent. In ADAS, Tesla claims to have implemented neural network functionality, based on the MobilEye EYEQ3 processor, in its autonomous driving control unit.

Key global players

Major players are still foreign tier-one companies. However, we think Chinese players may have an opportunity in voice recognition given their advantage in Chinese language.

Current supply chain structure

Major voice recognition companies currently supply auto OEMs through tier-one suppliers. We note voice technology providers are working with map suppliers and smart device system vendors since voice is an intuitive method to control vehicle while minimising driver distraction.

Exhibit 29: Deep learning in both ADAS and infotainment (HMI) areas

ADAS Infotainment and HMI-related

Camera-based vision system Voice recognition

Radar-based detection unit (LIDAR) Gesture recognition

Driver drowsiness to evaluate drivers' condition Eye tracking and driver monitoring

Sensor fusion ECU Virtual assistance and natural language interface

Source: BNP Paribas

Exhibit 30: Major AI supply chain companies

Company type Company name

IC and Component maker Nvdia, MobileEye, Xilinx, Intel, NXP, CEVA, Synopsys, Cadence

OEM and tier one system supplier Volvo, Daimler, BMW, Ford, Audi, Panasonic, Delphi, Bosch, Denso, Harman, iNago

Other technology companies Apple, Nuance, Sensory, VoiceBox, Google, Microsoft, Baidu, iFlytek

Source: BNP Paribas

Exhibit 31: Automotive voice recognition and virtual assistant value chain

Source: BNP Paribas

Voice Recognition technology

iFlytek, Nuance, Sensory

Tier 1 suppliers

Harman, Panasonic, and others

Auto OEMs

Audi, BMW, and others

Content provider

Map supplier

Baidu, Autonavi, and others

Voice through smart device

Siri, Cortana, and S voice

In-vehicle platform

Apple Carplay and Android Auto

Navigation software

NNG, Navinfo

Virtual Assistant Software

iNago

Cloud based in-vehicle assistant

New Value Chain Disruptor

VR Value Chain flow

20



3. High definition mapping

High definition maps are crucial to autonomous driving as they help cars understand their surrounding environment and enable key decision making. We think maps need to be very accurate, more so than for global positioning systems, so vehicles can understand surroundings in terms of various information, such as traffic lights, curb height, road barriers, potholes, width of roads, etc. We think high definition maps will drive SAE ‘level 3+’ autonomous driving.

An autonomous driving mapping system helps vehicles understand their surroundings in terms of these questions:

1 Where am I?

2 Where does the road go?

3 What is there around the corner if I continue along with the road?

4 How should I behave around that corner?

The difference between navigation maps and autonomous driving maps is that the former contains just basic information regarding lanes and road edges while the latter are more detailed with information down to land level accuracy and elevation. Ideally, mapping systems are connected to a smart vehicle’s ADAS, helping the car to understand the environment and assist in decision making.

Key global players

Major high quality mapping participants include HERE (Not listed), TomTom (TOMAF US, NR), Sanborn (Not listed), GeoDigital (Not listed), Tesla (TSLA US, NR), Toyota (7203 JP, NR), Mobileye (MBLY US, NR), Apple (AAPL, NR), Google (GOOG, NR), Uber (Not listed), and Baidu (BIDU US, BUY). We think China could be one of the biggest markets for automotive mapping, and could offer a better opportunity to local supply chains given local player advantage. We don't have clear market share information in the China auto map market but expect key players to be those with a stronger presence in mobile maps. As such, we think key players in China are Baidu, AutoNavi (Not listed) and Navinfo (002405 CH, NR).

Exhibit 32: High definition mapping is critical to autonomous driving vehicles

Source: BNP Paribas

Traffic lights

Space of the road

Traffic signs

Bridge and others

3D geometry

PotholesRoad barrier

21



Future trends in mapping system space – more and more applications

In the future, we believe high definition maps will not only facilitate autonomous driving development but also power other autonomous vehicles, like drones, as these maps cover details like sidewalks, buildings and electricity lines.

4. Human machine interface (HMI)

We believe the increasing importance of human machine interfaces (HMI) will be driven by autonomous driving technology as HMI, by definition, allows users to interact with machines’ control systems. In recent years, HMI has been adopted in automotive as a way for drivers to receive information and implement orders from/to vehicles. With the ongoing integration of advanced computer systems in vehicles, the HMI market should continue to expand with more complex designs.

Different input technologies in HMI

In recent years, with the ongoing integration of advanced computer systems into vehicles, HMI has become more complex. There are three major categories of interface input: 1) touchscreens; 2) physical controls, a combination of knobs/joysticks and separate displays; and 3) voice recognition.

Physical controls (knobs and joystick) are still the mainstream for automotive HMI inputs, but touchscreens and voice recognition are gaining penetration, and preferred

Exhibit 33: Mobile map share of times used per month Exhibit 34: Mobile map share of monthly active users

data as of July 2016 Source: iResearch

data as of July 2016 Source: iResearch

Exhibit 35: HMI concept

Source: BNP Paribas

Baidu69.4%

AutoNavi (Alibaba)

22.1%

Google 5.9%

Tencent2.7%

Baidu61.8%

AutoNavi (Alibaba)

30.0%

Google 4.8%

Tencent3.4%

HMI

Infotainment

Voice Control

ADAS I/ODisplay

NavigationTelematics

22



by drivers, according to Strategy Analytics. In China, only 9% of users prefer physical controls, compared to touchscreens (53%) and voice recognition (38%).

Exhibit 36: Pros and cons of automotive HMI inputs by category

Method Advantage Disadvantage

Physical Controls 1 Defined physical shape 2 Tactile feedback

Not suitable for complex tasks

Touchscreens 1 Flexible display of abundant content 2 Easy to master for all types of users

1 Poor for tasks involving long lists, such as searching for a phone contact

2 Driver distraction risks due to the lack of tactile feedback and the need for eye glance check

Voice Recognition 1 Potentially the safest without distraction 2 Well-suited for complex tasks such as

message entry, music playback, or radio tuning

Current speech recognition systems are not optimal

Handwriting Recognition

Good at deciphering input, especially attractive to market with non-Roman character sets, such as China

1 Not suitable for basic tasks, such as volume control

2 Requires large space in the centre console

Non-Touch Gestures

Very long learning curve for users

Eye Tracking Requires large amount of glance time, which causes driver distraction

Source: BNP Paribas

Future trends in HMI space

We expect both touchscreens and voice recognition to become mainstream solutions for HMI in 2020. We also expect penetration of multi-function controllers and heads-up displays to gradually pick up:

Touch panel penetration to rise. Touchscreens have been widely used inside vehicles and we think the trend will continue, driven by major OEMs’ designs (bigger screens with more use of capacitive touch).

Voice recognition is growing fast. We expect significant development in voice recognition technology as it is an intuitive method to control vehicles and helps reduce driver distraction. Biometrics-based access is convenient and personalised, and could be an ideal solution in the longer term.

Heads-up displays slowly pick up. We think heads-up display is a cost-efficient solution in the HMI space with less space requirement for installation. However, based on the current industry state, we don’t expect a big market for heads-up displays in the near future.

Exhibit 37: Global HMI technologies penetration Exhibit 38: Vehicles with resistive touch screen

Source: Frost & Sullivan Source: Frost & Sullivan

0 20 40 60 80 100

Air gesture

Digital instrument cluster

Multifunctional controller

Handwriting recognition

Voice Recognition

Head-up display

Touch screen

(%)

2021E 2016E 2015

0

10

20

30

40

50

60

70

80

90

100

North America Europe Asia

(%) 2015 2018 2021

23



Key global players

We think global tier one players in the HMI market remain international players with global platforms. Looking into the near future, we think companies will focus on integration with ADAS, vehicle to everything (V2E), and higher-end input systems. Major players are Continental, Harman, Bosch, Visteon, Delphi, Panasonic and Joyson (via Preh). Companies with strong voice recognition technologies such as Baidu may also provide integrated HMI systems in future.

5. Motion control

Advanced motion control is a critical part in autonomous driving vehicle as quality motion control can efficiently execute decisions made by the vehicle itself to avoid collision. Under an emergency braking situation, the system provides a strong braking force immediately to stop the vehicle and avoid collision. According to the European Union’s Commission, up to 1,100 fatal accidents could be avoided if all vehicles were equipped with such systems. To deliver better performance, we think electronic braking systems (EBS) will become much more important and a must-have system in autonomous vehicles. Key players in the market today are Bosch, Continental, TRW and Advics, based on market share data from Just-auto.

EBS needs a wide range of components, meaning opportunities for auto components suppliers. For example, Bosch introduced its iBooster in 2013, an electromechanical brake booster that provides support when the driver initiates braking. Bosch believes it will replace conventional vacuum-based boosters since it does not need a vacuum from the engine, it should meet all future requirements of electric drives and driver-assistance systems and it only draws electrical current during braking. We expect peers to follow with new products in the near future.

Exhibit 39: Vehicles with capacitive touch screen

Exhibit 40: Connected car outlook: global HUD supplier market share snapshot 2015

Source: Frost & Sullivan Source: Frost & Sullivan

0

10

20

30

40

50

60

70

80

90

100

North America Europe Asia

(%) 2015 2018 2021

Continental 45-50%

Denso10-15%

Nippon Seiki25-30%

Others20%

24



Exhibit 41: Global market share of electronic braking systems for passenger vehicles and light vehicles (volume)

Exhibit 42: Asia Pacific market share of electronic braking systems for passenger vehicles and light vehicles (volume)

Source: Just-auto Source: Just-auto

Exhibit 43: Structure of electronic braking system

Sources: Road Driver; BNP Paribas

0

10

20

30

40

50

60

70

80

90

100

2010 2011 2012 2013 2014 2015

(%) Bosch Continental TRW Advics Mando Others

0

10

20

30

40

50

60

70

80

90

100

2010 2011 2012 2013 2014 2015

(%) Bosch Continental Advics Mando Nissin Kogyo Others

Electronic Control Unit

Modulator Unit

Wheel Sensors

Wheel Sensors

Brake PedalBrake Booster

25



The opportunity for Chinese suppliers

In China, the penetration rate of auto electronics, including ESP, electronic brake assist, tyre pressure warning, and even navigation, is lower than the global average. Driven by potential Chinese regulations and safety concerns, we expect ADAS- and connected-car-system-related demand to grow in the next five years, and for this to motivate auto component makers. There are only a few Chinese companies in this field now, but replacement demand could encourage more to enter the market.

We expect opportunities in replacement demand especially from the component side and in products with relatively lower entry barriers like parking sensors and pressure-sensor-based tyre pressure warning systems. We think local suppliers will target local OEMs to start with given: 1) convenience and the long certification periods for foreign OEMs; and 2) pricing, since we expect a local vs international price gap of 20-30% based on our supply checks, although local suppliers have not really started mass production of ADAS- or connected-car-related products.

Exhibit 44: Vehicle stability control (ESP) penetration rate Exhibit 45: Electronic brake assist penetration rate


Exhibit 46: Pressure-sensor-based tyre pressure warning penetration rate

Exhibit 47: Navigation system penetration rate


0

10

20

30

40

50

60

70

2015 2016E 2017E 2018E 2019E 2020E 2021E 2022E 2023E

(%) China Global

0

10

20

30

40

50

60

70

80

90

100

2015 2016E 2017E 2018E 2019E 2020E 2021E 2022E 2023E

(%) China Global

0

10

20

30

40

50

60

70

80

90

100

2015 2016E 2017E 2018E 2019E 2020E 2021E 2022E 2023E

(%) China Global

0

10

20

30

40

50

60

2015 2016E 2017E 2018E 2019E 2020E 2021E 2022E 2023E

(%) China Global

26



Key players

We think our research is differentiated from other foreign broker reports by our identifying local Chinese suppliers through company visits and supply chain checks.

We see Greater China players as more like component suppliers than system providers (like the international players). We summarize our findings from conversations with various companies and relevant companies overleaf:

1 More players in sensor-related component area. China local companies have more exposure to the camera-based sensor field given clear market demand and better hardware component know-how, vs chipset or software fields. We think foreign system makers will eventually outsource the majority of hardware components to the Chinese supply chains given competitive prices and quality.

2 Limited material impact at this point. Based on our extensive company visits in China, we realize the current revenue contribution from ADAS or connected car systems is limited right now with most component suppliers only having low single digit percentage sales contributions expected by end-2016. Although many local players, such as APG and Toupu, expect to have entered mass production in 2017-18, we think there are still some uncertainties.

3 Better opportunity in areas where local players have local advantage. Currently, advanced auto systems in autonomous driving are only applied to luxury auto models with components dominated by international players. We don’t expect foreign OEMs to readily switch to local supplies due to the lack of track record and uncertain product quality. As such, we think the space where we see local advantage offers a bigger opportunity, like voice recognition (given Chinese language advantage) and domestic high-definition mapping.

27



Exhibit 48: ADAS major players – domestic versus international

Component type Greater China International

Environmental awareness

Camera Sunny Optical 舜宇光学 (2382 HK, BUY) O-Film 欧菲光 (002456 CH, NR), Hikvision 海康威视 (002415 CH, BUY), Wistron Neweb 啟碁科技 (6285 TT, BUY), Tung Thih 同致電子 (3552 TT, BUY), HiRain tech 恒润科技 (Not-listed), HSAE 航盛电子 (Not listed), SV Auto 德赛西威 (Not listed), Jingu 金固股份 (002488 CH, NR), Tatfook Tech 大富科技 (300134 CH, NR)

Mobileye (MBLY US, NR), Bosch (Not listed), Continental (CON GR, NR), Delphi (DLPH US, NR), TRW (Not listed), Autoliv (ALV US, NR) Valeo (FR FP, NR) Continental (CON GR, NR), Denso (6902 JP, NR) Valeo (FR FP, NR) Magna (MGA US, NR) Mando (204320 KS, NR) Nidec (6594 JP, BUY) Pasasonic (6752 JP, BUY)

Laser radar Great Star 巨星科技 (002444 CH, NR), Han's Laser 大族激光 (002008 CH)

Night view Protruly 保千里 (600074 CH, NR), SAT 飒特红外 (Not listed)

Navigation & routing

Baidu 百度 (BIDU US, BUY) Navinfo 四維圖新 (002405 CH, NR), AutoNavi 高德软件 (AMAP US, NR), Neusoft 東軟集團 (600718 CH, NR), Hi-target Navigation 中海達 (300177 CH)

HERE (Not listed), TomTom (TOMAF US, NR), Sanborn (Not listed), GeoDigital (Not listed), Tesla (TSLA US, NR), Toyota (7203 JP, NR), Mobileye (MBLY US, NR), Apple (AAPL, NR), Google (GOOG, NR), Uber (Not listed)

Motion control

APG 亞太股份 (002284 CH, NR), Vie Science & Technology 万安科技 (002590 CH, NR), Toupu 拓普集团 (601689 CH, NR) Wanxiang Qianchao 万向钱潮 (000559 CH, NR) Dongfeng Electronic Tech 东风电子 (600081 CH, NR)

Continental (CON GR, NR), TRW (Not listed), Bosch (Not listed) Mando (204320 KS, NR) Haldex (HLDX SS) WABCO (WBC US) Advics (Not listed) Knorr-Bremse (Not listed)

HMI

Joyson 均胜电子 (600699 CH, BUY), Baidu 百度 (BIDU US, BUY) Neusoft 東軟集團 (600718 CH, NR)

Continental (CON GR, NR), Harman (HAR US, NR), Bosch (Not listed), Visteon (VC US, NR), Delphi (DLPH US, NR), Panasonic (6752 JP, BUY)

Machine Deep learning & AI

iFlytek 科大讯飞 (002230 CH, HOLD), Baidu 百度 (BIDU US, BUY),

Panasonic (6752 JP, BUY), Delphi (DLPH US, NR), Bosch (Not listed), Denso (6902 JP, NR), Nvidia (NVDA US, NR), MobileEye (MBLY US, NR), Xilinx (XLNX US, NR) , Intel (INTC US, NR), NXP (NXPI US, NR), CEVA (CEVA US, NR), Synopsys (SNPS US, NR), Cadence (CDNS US, NR)

Source: BNP Paribas

28



Financial comparison

To analyse the financial performance of key autonomous driving players, we have compared margin profiles and ROE performance. Although auto software companies have much higher gross margins than auto component companies, their ROE are not necessarily higher due: 1) to higher operating expenses (including SG&A and R&D, which are critical for sales channels and product competiveness); and 2) to lower asset turnover ratios.

Exhibit 49: DuPont analysis on key autonomous driving players

Company BBG code ----------- Net margin ----------- --------- Asset turnover --------- --------- Leverage ratio --------- ---------------- ROE ----------------

2016E 2017E 2018E 2016E 2017E 2018E 2016E 2017E 2018E 2016E 2017E 2018E

(%) (%) (%) (x) (x) (x) (x) (x) (x) (%) (%) (%)

Software

iFlytek* 002230 CH 15.4 14.7 14.7 0.4 0.5 0.7 1.4 1.4 1.3 8.4 10.4 13.3

Neusoft 600718 CH 9.9 6.9 7.5 0.7 0.6 0.5 1.5 2.2 2.1 9.9 8.6 8.5

Navinfo 002405 CH 10.6 10.8 11.3 0.4 0.4 0.5 1.4 1.4 1.4 5.4 6.5 7.6

Baidu* BIDU US 13.5 18.6 22.5 0.4 0.4 0.5 2.2 2.1 1.9 11.4 17.2 20.7

Component

Toupu 601689 CH 14.7 14.9 15.3 0.8 0.9 0.9 1.3 1.3 1.3 15.4 16.6 17.3

APG 002284 CH 4.9 5.2 5.1 0.8 0.8 0.8 1.8 1.8 2.0 6.6 7.5 8.1

Great Star 002444 CH 16.8 17.5 18.4 0.6 0.6 0.6 1.2 1.2 1.2 11.9 12.4 13.1

Protruly 600074 CH 27.2 27.7 27.5 0.4 0.5 0.6 1.1 1.2 1.2 11.6 15.9 19.9

Jingu 002488 CH 2.2 2.6 3.0 0.6 0.8 0.8 3.2 4.0 5.2 4.5 8.4 12.9

Tung Thih* 3552 TT 13.5 13.5 13.5 1.1 1.2 1.4 2.2 2.2 2.1 33.4 36.5 39.6

Joyson* 600699 CH 3.6 4.2 4.4 1.5 1.1 1.2 1.5 1.8 1.8 8.3 8.4 9.5

Sources: Company data; Bloomberg consensus estimates, *BNP Paribas estimates

Exhibit 50: Gross margin comparison Exhibit 51: Operating margin comparison

Name BBG ticker 2011 2012 2013 2014 2015

(%) (%) (%) (%) (%)

Software iFlytek* 002230 CH 57.2 53.7 53.0 55.6 48.9

Baidu* BIDU US 67.0 64.3 55.8 53.4 49.8

Neusoft 600718 CH 29.2 30.9 28.1 28.3 31.0

Navinfo 002405 CH 84.6 80.4 78.7 81.2 76.2

Component Toupu 601689 CH 25.8 29.1 30.5 29.2 28.4

APG 002284 CH 15.4 14.8 16.5 16.5 16.4

Great Star 002444 CH 26.4 25.7 27.4 29.7 27.0

Protruly 600074 CH 10.8 0.0 3.7 42.8 41.7

Jingu 002488 CH 28.2 26.0 22.8 24.9 23.5

Tung Thih* 3552 TT 28.2 27.3 28.3 30.7 30.6

Joyson* 600699 CH 20.5 17.7 19.1 19.6 21.6

Name BBG ticker 2011 2012 2013 2014 2015

(%) (%) (%) (%) (%)

Software iFlytek* 002230 CH 20.9 16.5 18.0 16.5 11.3

Baidu* BIDU US 46.1 42.8 26.7 18.0 8.8

Neusoft 600718 CH 5.2 5.5 2.6 1.3 1.7

Navinfo 002405 CH 29.8 6.3 (0.8) 1.3 5.5

Component Toupu 601689 CH 13.9 14.3 17.2 15.7 13.8

APG 002284 CH 5.2 4.8 6.3 6.4 5.0

Great Star 002444 CH 15.5 13.9 15.4 19.3 15.6

Protruly 600074 CH 3.8 (9.5) (10.2) 30.9 25.2

Jingu 002488 CH 13.4 10.5 8.0 10.4 5.2

Tung Thih* 3552 TT 11.2 10.5 10.3 12.8 15.5

Joyson* 600699 CH 11.2 6.9 7.7 7.2 7.3

Source: Bloomberg, *BNP Paribas estimates Source: Bloomberg, *BNP Paribas estimates

29



Exhibit 52: Valuation comparison Name BBG code Area Price Mkt cap Earning growth ------- EPS ------ ------- P/E ------- ----- P/BV ----- ------ ROE ------

2016E 2017E 2016E 2017E 2016E 2017E 2016E 2017E 2016E 2017E

(LC) (USD m) (%) (%) (LC) (LC) (x) (x) (x) (x) (%) (%)

Greater China

iFlytek* 002230 CH Voice control HMI 28.4 5,517 28 34 0.41 0.55 68.8 51.5 5.6 5.2 8.4 10.4

Baidu* BIDU US Autonomous driving 182.2 63,097 (68) 38 28.31 50.28 42.9 24.2 4.6 3.8 11.4 17.2

Joyson* 600699 CH Auto HMI solution 33.7 3,478 73 63 0.71 1.15 47.7 29.2 2.6 2.4 8.3 8.4

Wistron Neweb* 6285 TT Vehicle communication module 98.0 1,072 2 32 6.07 7.99 16.2 12.3 2.6 2.4 16.5 20.2

Tung Thih* 3552 TT ADAS solutions 448.0 1,203 75 35 16.40 22.10 27.3 20.3 8.3 6.7 33.4 36.5

Vie S&T 002590 CH ABS, EBS 24.7 1,760 38 35 0.29 0.39 85.9 63.7 8.9 7.4 11.8 10.6

Neusoft 600718 CH In-vehicle infotainment 17.2 3,214 20 10 0.36 0.49 47.5 35.2 2.5 2.4 13.5 5.1

Navinfo 002405 CH Electronic mapping for navigation 22.3 3,551 76 31 0.20 0.29 109.7 76.5 2.0 5.0 5.4 6.5

Toupu 601689 CH Intelligent braking system 28.3 2,730 29 22 0.80 0.98 35.2 28.9 5.3 4.6 15.4 16.6

APG 002284 CH ABS, AEB, and ACC 16.4 1,810 27 21 0.27 0.33 61.8 50.0 4.5 4.2 6.6 7.6

Great Star 002444 CH Laser radar 17.4 2,803 27 18 0.57 0.67 30.5 25.8 3.5 3.2 11.9 12.4

Protruly 600074 CH Infrared radar 16.1 5,929 110 49 0.34 0.50 48.1 32.2 13.8 9.9 23.4 25.3

Jingu 002488 CH Camera sensor 16.9 1,286 57 95 0.15 0.28 116.0 60.5 5.3 5.0 4.5 8.4

Cub Elecparts 2231 TT TPMS 343.0 917 16 26 16.02 19.79 21.4 17.3 8.8 7.3 45.6 46.5

Greater China peers 37 36 54.1 37.7 5.6 4.9 15.6 16.7

Overseas

Nidec* 6594 JP Camera sensor and radars 9,323 27,342 6 18 327 384 28.5 24.3 3.3 3.0 12.9 13.1

Panasonic* 6752 JP ADAS sensor and deep learning 1,038 24,290 (12) 15 74 85 14.1 12.3 1.3 1.2 9.5 10.4

Denso 6902 JP ADAS sensors 4,015 31,200 9 14 355 407 11.3 9.9 0.9 0.8 7.9 8.6

Nuance NUAN US Voice control HMI 14.4 4,022 n.m. 5 1.51 1.64 9.5 8.8 2.2 2.2 20.9 23.8

Autoliv ALV US ADAS sensors 104.2 9,188 28 9 6.66 7.29 15.6 14.3 2.5 2.3 15.9 16.1

Delphi DLPH US ADAS sensors 69.7 19,006 13 8 5.99 6.68 11.6 10.4 6.2 4.8 59.6 49.6

Mobileye MBLY US Auto image sensing 41.7 9,135 146 52 0.71 1.07 58.5 39.0 13.6 9.9 28.1 31.6

Harman HAR US HMI 83.2 5,803 36 8 6.85 7.43 12.1 11.2 2.1 1.9 18.7 17.6

Visteon VC US ADAS sensors 70.8 2,408 (93) (2) 4.46 4.70 15.9 15.1 3.6 2.9 16.9 19.2

Continental CON GR ADAS sensors 184.5 41,406 14 7 15.59 16.62 11.8 11.1 2.5 2.1 21.6 19.6

Valeo FR FP ADAS sensors 51.1 13,656 19 14 3.71 4.22 13.8 12.1 3.0 2.6 22.5 22.2

Overseas peers 17 13 18.4 15.3 3.8 3.1 21.3 21.1

Priced 29 Sep 2016 Sources: Bloomberg consensus; *BNP Paribas estimates

Exhibit 53: FY17 P/E versus FY17 earnings growth – Greater China peers comparison

Exhibit 54: FY17 P/E versus FY17 earnings growth – Overseas peers comparison

Sources: Bloomberg; *BNP Paribas estimates Sources: Bloomberg; *BNP Paribas estimates

iFlytek*

Baidu*Joyson*

Wistron Neweb*

Tung Thih*

Vie S&T

Neusoft

Navinfo

Toupu

APG

Great StarProtruly

Jingu

Cub Elecparts

10

20

30

40

50

60

70

80

0 20 40 60 80 100 120

FY17 P/E (x)

FY17 earnings growth (%)

Nidec*

Panasonic*

DensoNuance

Autoliv

DelphiHarman

Continental Valeo

0

5

10

15

20

25

30

0 5 10 15 20

FY17 P/E (x)

FY17 earnings growth (%)

30



Wistron New Web (Laura Chen)

Earnings

Thanks to increasingly strong orders from auto-related business and resumed demand from US clients in smart-home-related products, which were initially delayed in 2Q16, we expect WNC to report 3Q16 revenue of TWD14.1b (+12.6% q-q; +2.9% y-y; vs the BBG consensus of TWD13.9b ). We expect WNC’s 3Q16 GM and OPM to remain stable at 12.8% and 4.2%, with net profit of TWD502m (+30.4% q-q; -17.0% y-y), mainly driven by more favourable product mix.

Looking into 4Q16, with a new project ramping up in smart-meter-related business in Europe and continuous momentum in smart-home- and auto-related business, we expect WNC to deliver revenue of TWD15.7b (+11.5% q-q; +9.6% y-y; vs BBG consensus of TWD15.4b) with a steady GM and OPM of 13.3% and 4.6% respectively, leading to net profit of TWD613m (+22.0% q-q; +20.2% y-y).

While WNC’s current revenue contribution from China is relatively modest, i.e. less than 10%, we believe the rising adoption of ADAS and the development in autonomous driving vehicle will be the potential long-term growth driver. WNC offers the product portfolio including automotive telematics connectivity modules, and 3G/4G telematics applications to smart meters and radars, and its major customers include SiriusXM and Harmanetc. We expect its auto related sales would see solid growth with CAGR 26% during 2016-2018 mainly driven by the increasing demand on ADAS application and wireless communication component for car, which will be one of the key hardware components for the success of autonomous driving vehicle.

Valuation and rating

We reiterate our BUY rating and lift our earnings for 2017E and 2018E by 6.3% and 10.2%, respectively, to reflect solid shipments in smart-home- and auto-related business. We believe WNC’s growth outlook remains intact both near and longer term, predominately from rising demand for connected cars, M2M and smart home applications. We raise our TP to TWD112 (from TWD92.23), based on 14x our 2017E EPS (multiple unchanged, but rolled over to our 2017 estimate). Downside risks to our call include 1) weaker-than-expected ADAS growth/penetration, 2) slower-than-expected regulation timeframe, and 3) unexpected market share loss to competitors.

Exhibit 55: Earnings revisions

---------------------- Prior ---------------------- ---------------------- Revised ---------------------- ---------------------- Change ----------------------

2016E 2017E 2018E 2016E 2017E 2018E 2016E 2017E 2018E

Year-end Dec-31 (TWD m) (TWD m) (TWD m) (TWD m) (TWD m) (TWD m) (%) (%) (%)

Revenue 55,595 66,836 71,944 55,704 66,674 75,529 0.2 (0.2) 5.0

Gross Profit 7,277 8,931 9,617 7,268 9,067 10,399 (0.1) 1.5 8.1

Operating Profit 2,547 3,099 3,353 2,539 3,255 3,694 (0.3) 5.0 10.2

Pre-tax Profit 2,554 3,127 3,379 2,545 3,323 3,723 (0.4) 6.3 10.2

Net Profit 2,034 2,501 2,704 2,018 2,659 2,978 (0.8) 6.3 10.2

EPS (TWD) 6.1 7.5 8.1 6.1 8.0 9.0 (0.8) 6.3 10.2

GPM (%) 13.1 13.4 13.4 13.0 13.6 13.8 OPM (%) 4.6 4.6 4.7 4.6 4.9 4.9 NPM (%) 3.7 3.7 3.8 3.6 4.0 3.9

Source: BNP Paribas estimates

31



Exhibit 56: Quarterly P&L

1Q16 2Q16 3Q16E 4Q16E 1Q17E 2Q17E 3Q17E 4Q17E 2016E 2017E 2018E

(TWD m) (TWD m) (TWD m) (TWD m) (TWD m) (TWD m) (TWD m) (TWD m) (TWD m) (TWD m) (TWD m)

Sales 13,296 12,544 14,121 15,743 14,983 14,530 17,245 19,916 55,704 66,674 75,529

Gross profit 1,770 1,597 1,808 2,093 1,998 1,895 2,369 2,804 7,268 9,067 10,399

Operating profit 694 528 594 723 694 674 834 1,052 2,539 3,255 3,694

Pretax profit 660 492 627 766 712 691 851 1,069 2,545 3,323 3,723

Net profit 518 385 502 613 569 553 681 855 2,018 2,659 2,978

EPS (TWD) 1.56 1.16 1.51 1.84 1.71 1.66 2.05 2.57 6.07 7.99 8.95

Margins (%) GM 13.3 12.7 12.8 13.3 13.3 13.0 13.7 14.1 13.0 13.6 13.8

OPM 5.2 4.2 4.2 4.6 4.6 4.6 4.8 5.3 4.6 4.9 4.9

NM 3.9 3.1 3.6 3.9 3.8 3.8 3.9 4.3 3.6 4.0 3.9

Change (y-y %) Sales 15.5 (0.3) 2.9 9.6 12.7 15.8 22.1 26.5 6.7 19.7 13.3

Gross profit 9.4 2.0 (2.1) 10.2 12.9 18.6 31.0 34.0 4.9 24.7 14.7

Operating profit 18.7 17.2 (6.1) 6.4 0.1 27.6 40.6 45.4 8.2 28.2 13.5

Pretax profit 12.7 (3.0) (20.5) 14.7 7.8 40.5 35.7 39.6 (0.2) 30.6 12.0

Net profit 12.3 (2.9) (17.0) 20.2 9.8 43.7 35.7 39.6 2.3 31.8 12.0

Change (q-q %) Sales (7.4) (5.7) 12.6 11.5 (4.8) (3.0) 18.7 15.5 Gross profit (6.8) (9.8) 13.2 15.8 (4.5) (5.2) 25.0 18.4 Operating profit 2.1 (23.9) 12.3 21.9 (4.0) (2.9) 23.7 26.1 Pretax profit (1.1) (25.5) 27.5 22.0 (7.1) (2.8) 23.2 25.5 Net profit 1.7 (25.7) 30.4 22.0 (7.1) (2.8) 23.2 25.5

Sources: Wistron Neweb; BNP Paribas estimates

32



Not-rated company visits

Neusoft (600718 CH, NR)

Software solution provider in China Established in 1991, Neusoft is a software solution provider in various segments including medical, auto electronics, transportation, banking, and energy. As of 2015, IT & system integration accounted for 79% of sales, then medical system (hardware only) 20%, and others 1%. According to Neusoft, it has been a software outsourcing supplier for many tier-one customers, including QCOM, Nokia, Toshiba, Nokia, Alpine, Harman and Denso. In Dec 2016, it announced it would raise up to RMB2,900m via private placement, but has not yet applied for government approval.

Auto electronics business to pick up gradually Neusoft offers a complete portfolio of in-car infotainment embedded systems, including in-car entertainment, navigation, telematics, communication and other control systems. Management expects to benefit from the auto electronics trend in the longer term as it has acted a software partner with global auto electronics system players, such as Alpine, Harman, and Denso, on the back of years of experience in IT and given attractive pricing. Currently, Neusoft targets its auto-electronics-related revenue contribution reaching 10-15% with a gross margin better than its corporate blended average. Neusoft is also aiming to build up direct business relationships with leading domestic OEMs.

Position in domestic medical market Neusoft has a presence in domestic medical system with more than a 50% allocation in the Chinese government social insurance area. Neusoft has built collaborative relationships with governments in different cities, and cooperates with different hospitals (more than 100 hospitals), which it believes strengthens its bargaining power. In the longer term, the company aims to leverage its position to integrate national medial projects and systems. Neosoft subsidiary Xikang Health’s management and service platform offers some medical devices and systems through its online platform and health centres in different cities. Neusoft manufactures digital medical products, such as CT, MRI, Digital X-ray and diagnostic ultrasound systems. Neosoft expects medical business to remain a key growth driver.

IPO plans and 1H16 earnings Neusoft plans to lower its holding at both Xikang and Neusoft Medical (hardware) from more than 70% to 30-40% by cooperating with strategic partners. Meanwhile, management does not rule out the possibility of initial public offerings (IPO) for both subsidiaries. It reported 1H16 earnings of RMB93m (EPS of RMB0.08), up 3.7% y-y. Revenue went up by 7.0% y-y to RMB3,382m in 1H16. Software and system integration revenues went up by 9.7% to RMB2,682m (GM at 30.6%), while medical system revenues declined by 2.9% y-y to RMB6,645m (GM at 47.8%).

33



Toupu (601689 CH, NR)

Auto part manufacturer in China Founded in 1983, Toupu is engaged in automobile parts manufacturing, including vibration control, suspension and mechatronic series. Its customers include international and China auto OEMs and automotive component suppliers. In 2015, shock absorber represented 54% of total sales, followed by internal functional components 38%, control arms 6%, and intelligent braking systems (IBS) 2%. In March 2015, the company listed on the Shanghai Stock Exchange.

Stable growth from shock absorber and internal functional component Shock absorber and internal functional components are two major businesses for Toupu, representing 92% of revenue in 2015. For shock absorber business, the company believes its advantages are: 1) lower prices than competitors such as Continental and TRW, and 2) in-time delivery to clients. For internal functional component business, its major products are internal vehicle rooves. Toupu thinks its product margin of c20%, which is higher than the peer average of 10-20% (according to Toupu), is mainly due to its in-house fibre supply.

Intelligent braking system – components for ADAS functions IBS is a component for adaptive cruise control, an ADAS function. IBS allows electronic control units to automatically control braking systems to avoid potential forward collisions. Toupu is developing software for its IBS and expects to start mass production from 2018. According to the company, the ASP of its IBS is RMB1,800-2,000, lower than Bosch’s RMB4,000. The company targets to cooperate with both local and foreign OEMs with initial sales contribution to start from 2017-18.

1H16 earnings review Toupu reported 1H16 earnings of RMB269m (EPS of RMB0.41), up 26.3% y-y. Revenues rose 15.8% y-y to RMB1,667m in 1H16. Shock absorber revenue rose 14.4% to RMB847m (GM at 34.9%), while internal function component revenue rose 11.9% y-y to RMB631m (GM at 24.5%). Despite a low contribution to 1H16 sales (3%), IBS revenues rose 127.8% y-y to RMB50m in 1H16.

Exhibit 57: Product portfolio - Toupu

Source: Company

Shock absorber Intelligent Braking System (IBS)Internal functional component Suspension

34



APG (002284 CH, NR)

Automotive braking system supplier in China Asia-Pacific Mechanical & Electronic Group (APG) is engaged in manufacturing braking system for automotives. Its main products include disk brakes, drum brakes, vacuum boosters, anti-lock braking system (ABS), and brake cylinders, which are mainly used in passenger vehicles, passenger buses and trucks. In 2009, the stock was listed on the Shenzhen Stock Exchange.

Brakes contribute more than 80% of revenue In 2015, disk brakes accounted for 67.8% of total revenues (GM at 16.6%), followed by drum brakes 15.3% (GM at 16.1%) and auto electronics control systems 2.7% (including ABS). Its brake product clients include BAIC, SAIC, FAW, Volkswagen, Nissan, Chery, and Changan. APG’s major competitors in the brake area are international brands such as TRW and Continental.

Entering the ADAS market APG targets to expand its business scope into the ADAS market by breaking into China local auto OEMs at the initial stage. Its ADAS products include ABS, autonomous emergency braking (AEB), and adaptive cruise control (ACC). The company expects small volume production from AEB and ACC in 2017, and mass production in 2018. APG is currently supplying ABS to local OEMs such as BAIC, JAC, JMC and Chery, mainly for passenger vehicles.

1H16 earnings review APG reported 1H16 earnings of RMB99m (EPS of RMB0.13), down 4.1% y-y. Revenues declined by 2.2% y-y to RMB1,688m in 1H16, dragged by price cuts from intensive market competition. APG guides for 9M16 earnings of RMB118-154m (up 0-30% y-y), mainly thanks to new business development.

Exhibit 58: Product portfolio - APG

Source: Company

Brakes Electronics component

35



Great Star (002444 CH, NR)

Global handheld tool supplier Founded in 1993, Great Star is a leading global hand tools and handheld electric tools manufacturer. Its major products include hand tools, power tools, and specialty tools (eg electrical tools and automotive tools), which are widely used in petroleum, chemical engineering, electronic, electricity power, automotive, and machinery. The company listed on the Shenzhen Stock Exchange in 2010.

No 5 position in global traditional tooling market Traditional tooling products represent 90%+ of Great Star’s revenues. According to the company, Great Star is the No 5 traditional supplier in the global market in terms of revenues, and it targets a top-three position in the next three years. Great Star currently has two business models for traditional tooling, its own brand and OEM/ODM. Its major marketing channels are international brands such as Home Depot, Lowe’s, King Fisher, Walmart, Carrefour, and Sears. Meanwhile, Great Star is competing with Stanley Black & Decker, Snap-on Tools, Apex, and Newell Rubbermaid in the international tooling market.

New driver from laser technology Great Star has launched new products, including electronic tooling and intelligent device, to expand its product portfolio. However, the contribution is limited currently with less than 5% revenues contribution in FY15. In July 2015, Great Star acquired Huada Kejie, a company specializing in opto-electromechanical integrated laser measurement and control, with a 65% stake in 2015. It believes that laser-related applications (3D radar and robot) could be a future driver and improve margins.

1H16 earnings review Great Star reported 1H16 earnings of RMB240m (EPS of RMB0.23), up 18.5% y-y. Revenues went up by 21.3% y-y to RMB1,567m in 1H16. Meanwhile, Great Star guides for 9M16 earnings of RMB461-544m (up 10-30% y-y), driven by continued growth of company’s major business and gross margin improvement.

Exhibit 59: Product portfolio – Great Star

Source: Company

Tools Safety Product

Storage

36



Appendix

Major ADAS functions

Adaptive cruise control (ACC) Adaptive cruise control (ACC) is an ADAS feature which allows driver’s cruise control system to adapt vehicle speed to the traffic environment. To measure and maintain safe distance from other vehicles on the road, ACC mainly uses sensors to gather information. Despite ACC system designs might vary between different manufacturers, the two major sensors are radar sensors and laser sensors. Due to the limit of lasers, laser sensors must be installed outside the vehicle. Also, laser sensors might not work well under adverse weather conditions and are not able to detect other vehicles which are dirty and non-reflective to the laser. In contrast to laser sensors, radar sensors are able to be equipped inside the vehicle.

As well as on-board sensors embedded in individual vehicles, there is an advanced technology named cooperative adaptive cruise control (CACC). CACC uses information gathered from vehicle to infrastructure (V2I) communication (infrastructure such as satellites and road side beacons) and vehicle to vehicle (V2V) communication to develop more optimized reactions towards traffic situations.

Lane departure warning (LDW) By gathering information from lane detection modules, lane departure warning (LDW) systems warn drivers when the vehicle is moving out of the road lane. This system is designed to decrease car incidents caused by driver error, distraction and drowsiness. Key components of the system are the sensors, which typically including video sensors, laser sensors and infrared sensors.

Blind-spot detection (BSD) Blind-spot detection (BSD) systems have a 360 degree of surrounding view to make sure every vehicle in the driver’s blind-spot is monitored. If the driver is changing lanes while there are vehicles in the blind spot or approaching rapidly from the rear, the system will warn by displaying a warning on the side mirror, or even vibrating the steering wheel. Sensors for the BSD system are equipped on the sides of the vehicle, usually around the external rear view mirrors or near the rear bumpers.

Forward collision warning (FCW) Forward collision warning (FCW) systems use sensors (radar, laser and camera) to detect a potential crash. Once a danger is detected, the system provides a warning to the driver or even takes action autonomously (eg braking at low vehicle speeds).

Night-vision (NV) Night-vision (NV) systems use cameras to increase driver’s perception under circumstances of adverse weather conditions or darkness. Two types of NV system are active and passive systems. Active systems use infrared light to illuminate the road ahead, while passive systems uses thermographic cameras to capture thermal radiation from other objects (eg humans or animals). Compared to active systems, passive systems have lower resolution images and poor performance in warmer environments. However, passive systems have longer detection ranges of more than 300 metres, while active systems only have ranges of less than 200 metres.

Park assist (PA) Park-assist (PA) systems are one of the most developed ADAS technologies. According to Frost & Sullivan, PA accounted for 61.7% of ADAS product shipments in 2012. The system combines cameras and sensors to judge the size of parking space and assist drivers to park through images showed on in-vehicle displays.

37



Major ADAS functions and technologies

Exhibit 60: Summary of ADAS functions and technologies

Product Definition

1) Sensing Camera A camera that detects and analyses information from inside and outside the vehicle to alert the driver and activate automatic control if necessary

A) Monocular Camera An automotive sensing camera equipped with a monocular camera (digital camera)

B) Stereo Camera A camera system using two lenses placed side by side for the recognition of three-dimensional spacial information. Stereo cameras

are used primarily for middle- and long-distance detection such as adaptive cruise control (ACC) and forward collision warning (FCW) in automotive sensing camera systems.

C) IR Camera An automotive sensing camera system using near- and far-infrared cameras. Infrared (IR) camera systems are used primarily in night vision systems (NVS) and night sensing systems (NSS).

Far-infrared camera An automotive sensing camera system using far-infrared cameras. Far-infrared camera systems are used primarily in NVS and NSS.

Because of the difference in wavelength from near-infrared cameras, far-infrared cameras can detect the heat emitted by objects. They are hence suited to detecting pedestrians and animals at least 200 metres away.

Near-infrared camera An automotive sensing camera system using near-infrared cameras. Near-infrared camera systems are used primarily in NVS and night detection system (NDS). Because their wavelength is closer to that of optical wavelength than that of far-infrared cameras, near-infrared cameras are suited to detecting objects at night in the short to medium distance within reach of near-infrared light. A near-infrared camera requires a near-infrared light projector along with a camera module.

View camera + sensing camera A view camera with an added sensing function.

Single A view camera with a single camera unit with an added sensing function.

Surround View A surround view system consisting of several cameras with an added sensing function.

2) Milliwave Radar A radar system that detects its surroundings in the short, medium to long distance using milliwave, sub-milliwave and other radars.

A) Milliwave Radar A radar system that detects its surroundings using milliwaves (30-300GHz). Milliwave radars for automotive sensing systems currently use the 76-77GHz band (described as 77GHz in this report) and will use the 79GHz band in the future. Their main application is medium-distance detection such as ACC, advanced emergency braking system (AEBS) and FCW.

B) Sub-Milliwave Radar A radar system that detects its surroundings using sub-milliwaves (10-30GHz). Sub-milliwaves radars for automotive sensing systems currently use the 24GHz band and will use the 25-26GHz band in the future. Their main application is short- to medium-distance detection such as cross traffic alerts (CTA), blind spot detection (BSD) and lane change assist (LCA).

3) Laser Radar A radar system that detects its surroundings mainly using near-infrared laser light.

A) Laser Radar Its main application is short- to medium-distance detection such as ACC and FCW.

B) Laser Finder Its main application is surround sensing for self-driving and self-parking cars.

4) Ultrasonic Sensor A sensor collects information outside the vehicle and notifies the driver, mainly for parking support.

Ultrasonic Sensor A sensor that detects objects and measures their distance by emitting ultrasound waves and receiving the reflected waves.

Its main application is parking support such as parking assist systems, but the product is likely to be installed in driving support systems as well in the future.

5) One Module Fusion A camera and a milliwave or laser radar are integrated into a single module/system for detecting the surroundings, mainly in the short, medium to long distance.

Milliwave + Monocular Camera A product integrating a monocular camera and a milliwave radar (77GHz) into a single module.

Laser Radar + Monocular Camera A product integrating a monocular camera and a laser radar into a single module.

6) View Camera A barebones camera that facilitates parking and other manoeuvres.

Sources: TSR; BNP Paribas

38



Major connected vehicle function

Exhibit 61: Examples of connected vehicle application Application Example Application Example

V2V Safety Emergency Electronic Brake Lights (EEBL) Mobility Advanced Traveller Information System

Forward Collision Warning (FCW) Intelligent Traffic Signal System (I-SIG)

Intersection Movement Assist (IMA) Signal Priority (transit, freight)

Left Turn Assist (LTA) Mobile Accessible Pedestrian Signal System (PED-SIG)

Blind Spot/Lane Change Warning (BSW/LCW) Emergency Vehicle Pre-emption (PREEMPT)

Do Not Pass Warning (DNPW) Dynamic Speed Harmonization (SPD-HARM)

Vehicle Turning Right in Front of Bus Warning (Transit) Queue Warning (Q-WARN)

V2I Safety Red Light Violation Warning Cooperative Adaptive Cruise Control (CACC)

Curve Speed Warning Incident Scene Pre-Arrival Staging Guidance for Emergency Responders (RESP-STG)

Stop Sign Gap Assist Incident Scene Work Zone Alerts for Drivers and Workers (INC-ZONE)

Spot Weather Impact Warning Emergency Communications and Evacuation (EVAC)

Reduced Speed/Work Zone Warning Connection Protection (T-CONNECT)

Pedestrian in Signalized Crosswalk Warning (Transit) Dynamic Transit Operations (T_DISP)

Road Weather Motorist Advisories and Warnings (MAW) Dynamic Ridesharing (D-RIDE)

Enhanced MDSS Freight-Specific Dynamic Travel Planning and Performance

Vehicle Data Translator (VDT) Drayage Optimization

Weather Response Traffic Information (WxTINFO) Smart Roadside Wireless Inspection

Environment Eco-Approach and Departure at Signalized Intersections Smart Truck Parking

Eco-Traffic Signal Timing Infrastructure Management

Weather Information

Eco-Traffic Signal Priority Winter Maintenance

Connected Eco-Driving Pothole Detection

Wireless Inductive/Resonance Charging Automated Mapping

Eco-Lanes Management Electronic Payment Tolling

Eco-Speed Harmonization Parking

Eco-Cooperative Adaptive Cruise Control Automotive Vehicle Diagnostics

Eco-Traveller Information Software Updates

Eco-Ramp Metering Agency Data Probe-based Pavement Maintenance

Low Emissions Zone Management Probe-enabled Traffic Monitoring

AFV Charging/Fuelling Information Vehicle Classification-based Traffic Studies

Eco-Smart Parking CV-enabled Turning Movement & Intersection Analysis

Dynamic Eco-Routing (light vehicle, transit, Freight) CV-enabled Origin-Destination Studies

Eco-ICM Decision Support System Work Zone Traveller Information

Sources: US Department of Transportation; Centre of Automotive Research

39

Ifl ytek 002230 CH BNP PARIBAS Jimmy H uang

THIS REPORT IS SOLELY FOR RECIPIENTS WHO ARE NON-CHINESE INVESTORS LOCATED OUTSIDE THE PEOPLE'S REPUBLIC OF CHINA ("PRC"). WITHOUT THE PRIOR WRITTEN CONSENT OF BNP PARIBAS, THE RECIPIENTS OF THIS REPORT SHALL NOT FURTHER DISTRIBUTE SUCH REPORT OR DISCLOSE ANY INFORMATION THEREIN TO ANY OTHER PERSON INCLUDING BUT NOT LIMITED TO ANY OTHER NON-CHINESE INVESTORS OR ANY PERSON IN PRC. 30 SEPTEMBER 2016 CHINA / SOFTWARE & SERVICES

IFLYTEK 002230 CH

The voice of China KEY STOCK DATA

Sources: FactSet estimates; BNP Paribas estimates

Initiating coverage at HOLD with a TP of RMB31We believe iFlytek is poised to benefit from China’s autonomousdriving and artificial intelligence (AI) trends, despite increasingcompetition in these areas. Its leading position in the voicerecognition market, particularly in China autos and education,should drive steady earnings growth. However, as its risk-reward profile is not attractive at this time, we initiate at HOLD.

Increasing voice recognition in auto spaceWe expect the voice recognition trend in auto HMI (humanmachine interface) to grow as voice is an intuitive method tocontrol vehicles and helps to reduce driver distraction. Eventually,biometrics-based access should also increase convenience andpersonalization, representing potential opportunities for iFlytek.

AI is a new opportunityThe Chinese government highlighted the importance of AI at therecent G20 summit in Hangzhou. We think iFlytek will benefitfrom wider adoption of AI, especially in education, given itsmachine perception capacity (e.g. voice/writing recognition).

Rich valuation from high entry barriers and strong growthOur TP of RMB31 is based on 57x FY17E P/E line with thehistorical trading avg. since 2011 as we expect an FY16-18earnings CAGR of 34%, vs 33% over FY10-15. Our DCFcrosscheck supports our TP. Trading at 52x FY17E P/E (vs.hist.58x), we initiate at HOLD and suggest investors wait for abetter entry point.

12m forward P/E versus net income

Sources: iFlytek; BNP Paribas estimates

Jimmy Huang [email protected] +852 2825 1124

HOLD TARGET PRICE RMB31.00

UP/DOWNSIDE +9.0%

CLOSE RMB28.43

HOW WE DIFFER FROM CONSENSUS MARKET RECS

TARGET PRICE (%) (16.2) POSITIVE 4

EPS 2016 (%) 0.7 NEUTRAL 0

EPS 2017 (%) 5.1 NEGATIVE 0

2015A 2016E 2017E 2018E

Revenue 2,501 3,524 4,944 6,906

Rec. net profit 425 544 727 1,016

Recurring EPS (RMB) 0.33 0.41 0.55 0.77

EPS growth (%) 4.8 24.8 33.7 39.7

Recurring P/E (x) 85.9 68.8 51.5 36.9

Dividend yield (%) 0.4 0.4 0.6 0.8

EV/EBITDA (x) 93.5 71.4 48.0 33.1

Price/book (x) 5.8 5.6 5.2 4.7

Net debt/Equity (%) (36.6) (26.9) (22.0) (19.7)

ROE (%) 8.5 8.4 10.4 13.3

Share price performance 1 Month 3 Month 12 Month

Absolute (%) (6.0) (12.2) 5.0

Relative to country (%) (4.0) (12.8) (9.5)

Next Results

Mkt cap (USD m) 5,617

3m avg daily turnover (USD m) 69.0

Free float (%) 63

Major shareholder China Mobile Communication (14%)

12m high/low (RMB) 40.90/24.19

3m historic vol. (%) 26.4

ADR ticker -

ADR closing price (USD) -

Issued shares (m) 1,318

YE Dec (RMB m)

October 2016

(15)

(4)

8

19

30

24

29

33

38

42Sep-15 Dec-15 Mar-16 Jun-16 Sep-16

Iflytek Rel to Shenzhen SE Composite(RMB) (%)

0

500

1,000

1,500

0

50

100

150

200

2011 2012 2013 2014 2015 2016 2017 2018

(RMB m)(x) 12m forward P/E (LHS) Net income (RHS)

2011 2012 2013 2014 2015 2016E 2017E 2018E

Historical 12m forward P/E avg.

40

Iflytek 002230 CH Jimmy Huang


Investment thesis

We initiate coverage of iFlytek at HOLD with a RMB31 target price, suggesting 9.0% upside based on 57x FY17E P/E.

Increasing adoption of voice control in auto space. We think iFlytek will stand to benefit from the increasing voice recognition trend in auto HMI area as we think voice is an intuitive method to control the vehicle and help to reduce driver distraction.

AI is an upcoming driver for iFlytek. China Gov. highlighted the importance of artificial intelligence (AI) at 2016 G20 and we think a wider adoption of AI, especially in education area, will offer new opportunity for iFlytek on the back of its machines perception, like voice/writing recognition ability.

Initiate at HOLD with RMB31 TP. Our TP of RMB31 is based on 57x FY17 P/E, which is in line with historical avg. since FY11 as our FY16-18E earnings CAGR of 34% is consistent with FY10-15E earnings CAGR of 33%. We crosscheck our TP with a DCF valuation (7.4% WACC). It is currently trading at 52x FY17 P/E, vs peers’ 54x average, and we see a less convincing risk-reward profile at >50x P/E, especially because we expect increasing competition from global peers, like Baidu, Apple, and Nuance.

Catalyst

1 Increasing adoption of voice control system as HMI for China smart car.

2 Better-than-expected growth in China voice recognition market.

3 More projects from China intelligent education teaching systems.

Risk to our call

Downside risks to our call are 1) worse-than-expected China voice recognition demand, 2) international/domestic players penetrating the China voice recognition market, and 3) worse-than-expected growth of China smart car market.

Upside risks to our call are 1) better-than-expected China voice recognition demand, 2) government policy support to China local voice recognition players, and 3) better-than-expected growth of China smart car market.

Company background Key assumptions

Founded in 1999, iFlytek is a Chinese software enterprise focusing on speech intelligence and artificial intelligence technology. The company also provides chip products, voice messaging software, and integration service for e-government systems.

2015 2016E 2017E (RMB m) (RMB m) (RMB m) Voice-based software 303 414 652 Education 761 1,066 1,582 Internet & Telecom 313 486 716 Public & Government 248 395 572 System Integration 757 978 1,200 Big data service 113 175 215 Total sales 2,501 3,524 4,944 Change (y-y %) 40.9 40.9 40.3 GM (%) 48.9 48.7 49.5

Key executives Age Joined Title

Liu Qingfeng 42 1999 Chairman Chen Tao 42 2007 Vice President Wu Xiaoru 43 2007 Vice President

http://www.iflytek.com/en/index.html Sources: Iflytek; BNP Paribas estimates

Principal activities (revenue, 2015) Earnings sensitivity

------------- FY16E ------------ ------------- FY17E ------------

(%) Bear Base Best Bear Base Best

Revenue growth 30.9 40.9 50.9 30.3 40.3 50.3

Gross margin 45.7 48.7 51.7 46.5 49.5 52.5

Net margin 9.8 15.4 21.4 9.0 14.7 20.8

Upside/downside (36.3) 0.0 38.8 (38.7) 0.0 41.3

Source: Company data Sources: Iflytek; BNP Paribas estimates

Event calendar

Date Event

Oct 2016 iFlytek FY3Q16 results

Nov 2016 Nuance FY4Q16 results

Feb 2016 Nuance FY1Q16 results

Mar 2016 iFlytek FY2016 results

Key earnings drivers include revenue growth and margins.

In our best case scenario, we forecast 2016 revenue growth/GM to be 50.9%/51.7%, which would increase our earnings estimates by 38.8%, all else being equal.

In our bear case scenario, we forecast 2016 revenue growth/GM to be 30.9%/45.7%, which would decrease our earnings estimates by 36.3%, all else being equal.

Voice-based software - 12.1%

Education - 30.4%

Internet & Telecom - 12.5%

Public & Government - 9.9%

System Integration - 30.3%

Big data service - 4.5%

Others - 0.3%

41



Executive summary

We initiate coverage on iFlytek with a HOLD rating and a target price of RMB31, implying 9.0% upside from current levels. Founded by the doctors from University Science and Technology of China, iFlytek is a leading voice recognition expert in China with applications mainly including automotive and education markets. We think it should benefit from the development of both autonomous diving as well as artificial intelligence (AI) in China. We appreciate its leading position in China voice recognition market which will drive steady earnings growth of 34% earnings CAGR in 2016-18. However, we find its risk-reward profile less convincing at 50+ P/E. We suggest investors await a better entry point.

Key beneficiary from autonomous driving development

We see iFlytek as one of the major beneficiaries of the continued autonomous vehicle development as voice input and speech recognition appears to be an intuitive method to control the vehicle, which can also help to reduce driver distraction. Currently, iFlytek targets to penetrate into more than 100 auto models in FY16, vs 30 models in FY15, supported by its superior quality and accuracy. Its end customers in auto space include both global and China local OEMs, like Porsche, BMW, Volkswagen, Honda, SAIC, FAW, Changan, Geely, and Chery, and Didi Dache, a leading mobile platform taxi-calling application in China. Meanwhile, we believe biometrics-based voice access can also increase the convenience and personalization, which could be ideal solution in the long term.

Artificial intelligence (AI) provides longer term opportunity

China’s President Mr. Xi highlighted the importance of artificial intelligence (AI) during the week of 2016 G20 in Hangzhou, which suggests a directional trend in China. As both voice and image recognitions are the major applications today, we believe iFlytek can ride on the ongoing AI development, on the back of its deep language recognition, writings recognition, logical reasoning, and self-learning AI systems. Notably, iFlytek has invested in AI-related technology in education and Gov. court business with meaningful impact to start in FY17 at earliest, in our view. We expect both education and voice-based software to become key driver in FY16-18, which will represent 12-14% and 30-35% of total sales, respectively.

Waiting for a better entry point on rich valuation

Our TP of RMB31 is based on 57x FY17E P/E, which is in line with historical avg. since FY11 as our FY16-18 earnings CAGR of 34% is consistent with FY10-15 earnings CAGR of 33%. On the other hand, our TP of RMB31 implies 1.7x FY17 PEG, which is also in line with global peer Nuance’s current FY17 PEG level. We also use DCF method to crosscheck our TP; this gives a valuation of RMB31/share, based on a WACC of 7.4%. It is currently trading at 52x FY17 P/E, vs peers’ 54x average, and we see a less convincing risk-reward profile at >50x P/E, especially we expect an increasing competition from global peers, like Baidu, Apple, and Nuance. We suggest waiting for a better entry point.

Exhibit 1: Historical 12m forward P/E Exhibit 2: FY17 P/E versus earnings CAGR FY16-18E

Sources: Bloomberg; BNP Paribas estimates Sources: Bloomberg; BNP Paribas estimates

30

50

70

90

110

130

150

Sep-13 Sep-14 Sep-15 Sep-16

(x)

-1SD

Avg.

+1SD

+2SD

+3SD

iFlytek*

Kingsoft

YonyouDigiwin

Navinfo Thunder Software

Wisesoft

Hikvision*

Joyson*

10

20

30

40

50

60

70

80

90

100

110

0 50 100 150

FY17 P/E (x)

Earnings CAGR FY16-18E (%)

42



Potential share price catalysts

Better than expected earnings performance in 2016-17

We expect the FY16-17 bottom line to up by 28%/34% y-y, supported by steady revenue growth of 41%/40% y-y in FY16-17, with gross margin to expand slightly. Potential upside will come from the better than expected gross margin, especially after its AI related business (higher gross margin) starts generating revenue.

International award announcement

iFlytek has proved its leading voice and speech recognition by winning 4th CHiME (Computational Hearing in Multisource Environments) competition, which is held by The French Research Institute for Computer Science and Automatic Control, University of Sheffield, and Mitsubishi Electric Research Laboratories. The participants in the competition include Google, Cambridge University, Carnegie Mellon University and other globally renowned research institutions. We think any further similar announcement would reinforce market confidence on its market position and could provide a near-term stock price catalyst.

Better progress in auto models

Management expects to step into 100 auto models in 2016, from 30 models in 2015, and we think any better than expected voice recognition penetration rate in auto could be positive for overall sentiment.

Where we differ from the Street

We are the only broker on the street to have relatively cautious view on iFlytek, on the combination of its earnings growth as well as current valuation.

We identify voice control HMI as an important growth driver to iFlytek’s voice-based related product, which is able to penetrate tier 1 Auto OEMs in the future.

We expect a slightly higher gross margin than Bloomberg consensus, driven by a favourable product mix as voice-based softwares deliver a higher margin profile.

Exhibit 3: BNPP vs street consensus

------------------------- 2016E ------------------------- ------------------------- 2017E ------------------------- ------------------------- 2018E -------------------------

BNPP Street Diff. BNPP Street Diff. BNPP Street Diff.

(RMB m) (RMB m) (%) (RMB m) (RMB m) (%) (RMB m) (RMB m) (%)

Sales 3,524 3,501 1 4,944 4,919 1 6,906 6,863 1

Gross profit 1,717 1,731 (1) 2,449 2,432 1 3,445 3,423 1

Operating profit 412 432 (5) 651 613 6 980 853 15

Pre-tax profit 609 591 3 811 789 3 1,140 1,054 8

Net profit 544 532 2 727 710 2 1,016 948 7

Margins (%)

Gross margin 48.7 49.4 49.5 49.4 49.9 49.9 Operating margin 11.7 12.3 13.2 12.5 14.2 12.4

Pre-tax margin 17.3 16.9 16.4 16.0 16.5 15.4

Net margin 15.4 15.2 14.7 14.4 14.7 13.8

Sources: Bloomberg consensus; BNP Paribas estimates

43



Market leader in China voice recognition market

iFlytek has been a market leader in China voice technology area in various sectors, including education, automotive, and robotic areas. In 2015, it is ranked the No.1 player with market share of 44% in China speech recognition market, followed by Baidu 28%, Apple 7%, and Nuance 3%, according to Speech Industry Alliance of China (SIAC). SAIC expects China intelligent speech market to grow fast with 55% CAGR over 2015-17E as speech technology penetration remains low in China.

While the global market is dominated by US companies, such as Nuance, Google, Apple, Microsoft, and IBM, we believe China local players have a clear advantage in China market, in terms of Chinese database and Chinese speech recognition. Therefore, we are not surprised by the fact that Chinese local players together grab more than 70% of voice recognition market share in China as, generally, the nature of Chinese language creates an entry barrier to overseas companies.

A broader applications today but the entry barrier is very high

We have seen voice recognition moving away from traditional application like smartphone to various new territories including automotive, education, or robot. In terms of function, we think the voice recognition can provide not only the pure data input but also different infotainment services to support functions such as voice biometrics, virtual assistant, advanced machine learning, and so on.

Voice recognition market has an extremely high entry barrier in every country, especially for a foreign player. We think the voice recognition generally has a relatively low level of accuracy as the quality is highly influenced by factors such as

Exhibit 4: China intelligent speech market size

Source: SIAC

Exhibit 5: China speech recognition market share (2015) Exhibit 6: Global speech recognition market share (2015)

Source: SIAC Source: SIAC

0

20

40

60

80

100

120

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

2011 2012 2013 2014 2015 2016E 2017E

(y-y %)(USD m) Market size (LHS) Growth (RHS)

iFlytek44%

Baidu28%

Apple7%

Nuance3%

Xiaoi3%

Jietong Huasheng

2%

Haitian Ruisheng

2%Zongke Xinli

1%

Others10%

Nuance32%

Google28%

Apple 15%

Microsoft8%

iFlytek5%

IBM4%

Others8%

44



background noise, mood and physical conditions. In other words, the algorithm must be able to recognize the voice accurately under different environments; especially some places are noisy, like inside a fast moving vehicle. As such, market players have to spend a large amount of R&D expenditure to develop recognition technology, based on a sizeable database, language decoder, or core spoken language model. Due to a high requirement on R&D, voice recognition players were usually initiated from research institution. For example, Nuance, largest speech recognition player in global market, was originated from the Speech Technology and Research Laboratory of SRI, a world-leading research organization in speech technology area. On the other hand, iFlytek was also founded by the students of the University Science and Technology of China (中国科学技术大学).

Who are the key clients?

iFlytek has a diversified customer portfolio from various industries with top 5 customers together accounting for 25-30% of total sales in 2012-15. In education area, its major clients are educational institutions in Guangdong, Jiansu, and other provinces. In auto area, iFlytek cooperates with not only auto components suppliers (Continental and Delphi) but also auto OMEs such as BMW, Volkswagen, Changan, Great Wall and many others. Its products are also widely used in financial institutions, such as Bank of Communication, ICBC, and Ping An Insurance. In telecom space, three of China’s telecom operators are using iFlytek’s service for its voice-related functions. We believe the well-diversified client portfolio is beneficial for iFlytek to deliver a stable growth.

Exhibit 7: iFlytek - major client and partners Area iFlytek major clients and partners

Education Educational institutions in Guangdong, Jiangsu, Guangxi, Chongqing, and etc

Auto OEMs Porsche, BMW, Volkswagen, Honda, Mazda, Lexus, SAIC, FAW, Changan, Great Wall, Geely, Chery, JAC, GAC, and etc

Auto components Continental, Delphi, and etc

Telecom China Mobile, China Unicom, and China Telecom

Tech Intel, Huawei, Xiaomi, Tencent

Smart hardware Yunji Technology, IU House, JD.com

Financial Institution Bank of Communication, Ping An Insurance, Merchants Bank, ICBC

Broadcasting Tianjin Television & Broadcasting Group, Guangdong Television & Broadcasting Group, and etc

Government China Government agencies in different cities

TV vendors Konka, Changhong, Skyworth, Hisense

Sources: iFlytek; BNP Paribas

45



Voice recognition in auto space

We believe the autonomous driving demand is one of the key drivers to iFlytek in the long term because voice recognition plays a more important role in auto market.

One of the preferred control solutions in HMI

We expect iFlytek to benefit from the increasing demand of advanced human machine interface (HMI), especially voice control appears to be one of the key input methods. HMI, by definition, is a device that allows users to interact with machines’ control system. In recent years, HMI has been adopted in automotive as a way for drivers to receive information and command orders from/to the vehicles. We expect the automotive HMI market to expand, driven by the growing popularity of intelligent car (smart car) and infotainment demand.

We see interfaces have become more complex as a result of an ongoing integration of advanced computer systems into vehicles. There are three major categories for interface inputs: 1) through touchscreens; 2) through physical controls, a combination of knobs/joysticks and a separate display; and 3) through voice recognition. Based on Frost & Sullivan data, we think voice control is pick up fast, together with touch panel screen, in automotive HMI applications, on the back of (1) safe without distraction, and (2) improving voice recognition technology today. According to Frost & Sullivan, touch screen and voice recognition will become mainstream input for auto HMI, with penetration rate reaching 70%+, respectively.

Exhibit 8: Key development in voice recognition

Source: BNP Paribas

Voice Biometrics

Deep Speech

Embedded Natural Language Processing

Real-time Translation

Noise Cancellation

Voice Analytics to improve Voice

AI-enabled Virtual Assistants

46



Exhibit 9: Pros and cons for automotive HMI inputs by categories

Method Advantage Disadvantage

Physical Controls 1) Defined physical shape 2) Tactile feedback

Not suitable for complex tasks

Touchscreens 1) Flexible display of abundant content 2) Easy to master for all types of users

1) Poor for tasks involving long lists, such as searching for a phone contact 2) Driver distraction risks due to the lack of tactile feedback and the need for eye glance check

Voice Recognition 1) Potentially to be safest without distraction 2) Well-suited for complex tasks such as message entry, music playback, or radio tuning

Current speech recognition systems are not optimal enough

Handwriting Recognition

Good at deciphering input, especially attractive to market with non-Roman character sets, such as China

1) Not suitable for basic tasks, such as volume control 2) Requires large space in the center console

Non-Touch Gestures

Very long learning curve for users

Eye Tracking Requires large amount of glance times, which caused driver distraction

Sources: Strategy Analytics; BNP Paribas

We believe iFlytek could continue to improve its accuracy, supported by experienced speech synthesis and recognition technology, especially for Chinese voice recognition area. In China, 38% of auto consumers prefer speech recognition as HMI input, only behind USA’s 39% and Italy’s 40% in global market. As such, we think China could be a fast growing market for automotive voice control system, which offer good opportunity to local voice experts. Currently, iFlytek is cooperating with both global and domestic Auto OEMs, such as Porsche, BMW, Volkswagen, Honda, Mazda, Lexus, SAIC, FAW, Changan, Great Wall, Geely, Chery, JAC, GAC, and others, to work on in-vehicle voice control system.

Exhibit 10: HMI penetration rate by categories Exhibit 11: Consumer preference in HMI input

Source: Frost & Sullivan Sources: Strategy Analytics, ACI

0 50 100

Air gesture

Digital instrumentcluster

Multifunctional controller

Handwriting recognition

Voice Recognition

Head-up display

Touch screen

(%)

2021E 2016E 2015

43 48 45 5337

53

1822 27

20

239

39 29 28 2740 38

0

20

40

60

80

100

US UK France Germany Italy China

(%)

Speech recognitionRotary dial and/or hard buttonsTouchscreen

47



What is beyond voice input function?

Eventually, biometrics-based access can also increase the convenience and personalization, which could be ideal solution in the longer term, compared to other biological recognition technologies, such as fingerprint, palm print, and iris identification, voiceprint recognition delivers high security performance. Moreover, the technology requires any special except for a telephone and a microphone. As such, voiceprint recognition is the only contactless biological recognition technology which can be used for remote control over a telephone channel.

Exhibit 12: Automotive Voice Recognition value chain

Sources: Frost & Sullivan; BNP Paribas

Exhibit 13: Global Voice Recognition trends in automotive area

Trends Technology Provider Current Field of Application Expected Entry in Automotive industry

Voice biometrics Nuance, Sensory, and VoiceBox Banks 2018

Smartphone voice assistant Apple Siri, Cortana, S Voice, Google Now, iFlytek, and Maluuba

Smartphones Currently available

Voice recognition with wearables Microsoft Automotive and health/fitness Volvo with Microsoft

Virtual assistant iFlytek, Nuance and iNago Smartphones Expected by 2020

Noise cancellation Harman Automotive Commercially available

Machine learning iFlytek, Baidu, Google, VoiceBox, Nuance, Sensory, and iNago

Automotive and smart homes Expected by 2020

Sources: Frost & Sullivan; BNP Paribas

Voice Recognition technology

iFlytek, Nuance, Sensory

Tier 1 suppliers

Harman, Panasonic, and others

Auto OEMs

Audi, BMW, and others

Content provider

Map supplier

Baidu, Autonavi, and others

Voice through smart device

Siri, Cortana, and S voice

In-vehicle platform

Apple Carplay and Android Auto

Navigation software

NNG, Navinfo

Virtual Assistant Software

iNago

Cloud based in-vehicle assistant

New Value Chain Disruptor

VR Value Chain flow

48



Exhibit 14: North America automotive voice recognition market (2015)

OEM Embedded

voice Natural voice

Voice engine Voice through Smartphone Integrator Biometrics R&D Vehicle model

Audi Nuance Carplay, Android auto Harman Voice and facial Standard on 2016 models

BMW Nuance Carplay, Android auto Harman Voice biometrics Expected by 2016/2017

Chrysler Voicebox Carplay, Android auto Panasonic Facial and eye tracking Expected by 2016/2017

Chevrolet Nuance Carplay, Android auto Panasonic Not available Standard on 2016 models

Cadillac Nuance Carplay, Android auto Bosch Voice recognition Standard on 2016 models

Ford Nuance Carplay, Android auto Panasonic Eye tracking, facial recognition Standard on 2017 models

Honda Nuance Carplay, Android auto Denso Not available 2016 Accord and Civic

Hyundai Nuance Carplay, Android auto N/A Voice biometrics Sonata and Elantra

Mercedes-Benz Nuance Carplay, Android auto Harman Facial and voice CLA, CLS, E-class, and GLA

Toyota Voicebox No plans Denso Not available No timeline

Mazda Voicebox Not available Elektrobit Not available No timeline

Source: Frost & Sullivan

49



Long term opportunity from Artificial Intelligence (AI)

Generally, the concept of Artificial Intelligence (including machine learning and deep learning) can be dated back to 1980s. AI attempts to simulate the function of the human brain by using electronics and software algorithms. In Artificial Intelligence (AI) research scope, there are four major areas: 1) symbolic intelligence, 2) computing intelligence, 3) machine learning, and 4) machine perception. iFlytek’s advantage is at the machine perception side, which includes image identification, natural language processing, voice recognition and machine vision. On the back of its know-how in speech recognition space, iFlytek could develop a more advanced research in machine perception area. We think AI will provide a longer term opportunity to iFlytek. According to Global Times (click here), China’s President Mr. Xi highlighted the importance of both artificial intelligence (AI) and virtual reality (AR) during the G20 in Hangzhou in 2016, which should drive the continued market focus on this field, in our view.

In 2014, iFlytek launched China’s first artificial intelligence (AI) project, “Super Brain” (讯飞超脑). The “Super Brain” project emphasizes on deep language recognition, logical reasoning, and self-learning AI systems, based on company’s previous research on Chinese voice recognition. The systems will be applied in various areas, such as education, customer service, HMI, and medical. To facilitate the development of its AI projects, iFlytek will invest RMB352m to build research center and procure additional equipment for “Super Brain” project. Currently, management indicates the material impact remains quite limited in FY16-17 but expects the potential sales contribution to be significant in the longer run.

AI on automotive

We have observed that AI technology is used in auto area, especially ADAS functions and infotainment HMI. For example, one of the HMI functions, speech recognition, relies on algorithms based on neural networks running in the cloud. We expect a wider adoption of Artificial Intelligence in high-end auto market as tier 1 OEMs are attempting to improve its driver experience. According to IHS, deep learning based infotainment system will see a 32% CAGR over 2016-2025E in terms of shipment, which is mainly driven by speech recognition growth.

Exhibit 15: Research area of Artificial Intelligence

Source: BNP Paribas

Artificial Intelligence

Symbolic Intelligence

Graph searchingSymbolic learning

Uncertain reasoningAutomated reasoning

Knowledge engineering

Machine perception

Image identificationNatural language processing

Voice recognitionMachine vision

Machine Learning

Deep learningStatistical learningMatrix recognitionInductive learning

Computing Intelligence

Ant colony algorithmsArtificial immune systemEvolution Computation

Neural computingiFlytek’s area

50

http://www.globaltimes.cn/content/1004489.shtml



Exhibit 16: Deep learning-based infotainment systems Exhibit 17: Automotive application of AI

Applications Functions

Camera-based machine vision systems Object-marking recognition, context evaluation, action prediction

Radar-based detection units (LIDAR) Object recognition

Driver drowsiness Evaluation of driver´s condition

Sensor fusion ECU Context evaluation and action prediction

Human machine interface Speech recognition, gesture recognition, eye tracking & driver monitoring, virtual assistance and natural language interface

Source: IHS Source: BNP Paribas estimates

0

20

40

60

80

100

120

140

2015 2017E 2019E 2021E 2023E 2025E

(m units) Speech Recognition Gesture RecognitionVirtual Assistance

51



What is iFlytek’s core technology?

We think iFlytek’s core technology is based on four key engines and it applies those technologies to different industry and applications.

Speech Synthesis Engine 语音合成引擎 Speech synthesis technology, which is also known as the Text to Speech (TTS) technology, is the combination of multi technologies in different areas including acoustics, linguistics, digital signal processing, and computer sciences. The main function of speech synthesis technology is to convert text into intelligible voices, in other words, let machines talk more like human beings. With over 20-year experience in speech synthesis market, iFlytek holds a leading position in China market and occupies over 70% of market share.

Speech Recognition Engine 语音识别引擎 iFlytek’s Auto Speech Recognition (ASR) technology targets to equip a computer with voice input so that the machine is able to recognize the speech of human beings and then extract text information from the speech. As such, the ASR technology enables human-machine communication and interaction through a more natural and convenient ways. The technology plays an important role in intelligent computer systems. iFlytek has launched its speech recognition engine “InterReco” for China speech recognition market. Currently, iFlytek has gained over 80% market share of Chinese speech recognition market, especially in telecommunications, financial, energy, transportation, and government institute.

Voiceprint Recognition Engine 声纹识别引擎 Voiceprint recognition is a technology which automatically determines the identity of the speaker based on voice parameters in the voice waveform reflecting physiological and behavioural characteristics of the speaker. iFlytek has launched its speaker verifier product “InterVeri”, a professional voiceprint recognition engine to fulfil market demand in the security field. To elaborate its function: 1) the engine compares the voice of speakers with the voiceprints registered in the database, 2) checks and verifies the speaker’s identity, and 3) determines whether the voice matches the voiceprint of the speaker or not.

Voice Cloud 讯飞开放平台 To leverage speech technology in more mobile Internet applications, it launched iFlytek Voice Cloud in 2010. Using cloud computing technologies to process massive data, iFlytek voice cloud is a key intelligent speech technology platform designed for mobile Internet applications. Based on this voice platform, iFlytek has launched a series of applications for demonstration purpose, including iFlyDictation, iFLYTEK Voice Touch, and Lingxi Voice Assistant. In 2012, iFlytek has launched a new generation of open platform for the voice cloud, attracting over 10,000 partners including Huawei, Lenovo, Changhong, SAIC Motor, Sina, and Xiaomi.

Exhibit 18: iFlytek’s mobile application

Source: iFlytek

iFlytek Input Lingxi Voice Assistant iFlyDictation

Icon

Chinese name 讯飞输入法灵犀讯飞语音输入

Feature

iFlytek Input is an intelligent inputmethod to improve input experience onmobile devices through incorporating thespeech, handwriting, and Pinyin inputmethods.

As an intelligent voice assistant jointlylaunched by China Mobile and iFlytek,Lingxi serves as a voice secretary, whichenables users to make a voice call, sendmessages, check weather conditions,and even buy lotteries.

iFlyDictation is a speech input assistantsoftware that can hear and speak wordsbased on cloud computing. The softwaresupports speech input, speech-to-textconversion, and text sending by usingEmail, MicroBlog, WeChat and QQ.

52



Private placement approved by China authority

On 18 May 2016, iFlytek announced a private placement proposal to issue a total of 23.9m new shares, of which 12.9m shares are for acquisition Lezhixing (乐知行), an education software company in Beijing, and 11m shares for iFlytek’s operating capital demand. The potential dilution of current outstanding shares is around 2%. On 28 July 2016, the proposal has been approved by China Securities Regulatory Commission. We have factored in additional shares into our model as we believe the company will execute the private placement in the near term. On 10 Sep 2016, iFlytek officially announced that the acquisition of Lezhixing will be executed through cash (RMB143m) and new issue stocks (12.9m shares). However, the deal is still pending for final approval from China Securities Regulatory Commission and we haven’t factored in Lezhixing financials due to uncertain consolidation date.

53



Business overview

iFlytek has six major business lines, which are voice-based software, education, internet & telecom, public & government, system integration, and big data service. Among the six business lines, education and system integration both accounted for 30% of total revenues in FY15, respectively, followed by internet & telecom 13%, voice-based software 12%, public & government 10%, and big data service 5%.

1 Voice-based software (12% of total sales in FY15)

Voice-based software is the key engine for voice application system and product, such as telecom-level voice platform and embedded voice software. Its clients are voice application developers in various industries, such as telcos, robot manufacturing, car makers and financial institutions. iFlytek provides voice-based software to its clients mainly through licensing. Notably, the company does not

Exhibit 19: iFlytek revenues by business Exhibit 20: Revenues mix by business

Sources: iFlytek; BNP Paribas estimates Sources: iFlytek; BNP Paribas estimates

Exhibit 21: Business overview by business

Segment 2015 2016E 2017E 2018E

Revenues (RMB m)

Voice-based software 303 414 652 991

Education 761 1,066 1,582 2,419

Internet & Telecom 313 486 716 1,006

Public & Government 248 395 572 802

System Integration 757 978 1,200 1,423

Big data service 113 175 215 258

Revenues mix (%) Voice-based software 12 12 13 14

Education 30 30 32 35

Internet & Telecom 13 14 14 15

Public & Government 10 11 12 12

System Integration 30 28 24 21

Big data service 5 5 4 4

Growth (y-y %) Voice-based software 23.2 36.7 57.4 51.9

Education 15.0 40.2 48.4 52.9

Internet & Telecom 52.2 55.4 47.2 40.6

Public & Government 63.0 59.3 44.8 40.4

System Integration 61.0 29.3 22.7 18.5

Big data service na 55.6 22.4 20.0


0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

2011 2012 2013 2014 2015 2016E 2017E 2018E

(RMB m)Voice-based software EducationInternet & Telecom Public & GovernmentSystem Integration Big data serviceOthers

0

20

40

60

80

100

2013 2014 2015 2016E 2017E 2018E

(%)

Voice-based software EducationInternet & Telecom Public & GovernmentSystem Integration Big data serviceOthers

54



transfer the ownership of voice technology to its client. iFlytek recognizes its auto HMI related products in this business line. Currently, the company provides service modules to tier 2 auto components supplier. In the future, iFlytek targets to penetrate into Auto OEMs directly. In addition, the company supplies software to up to 5,000 small robot players in China.

2 Education (30% of total sales in FY15)

iFlytek provides education products including 1) intelligent speech teaching aid system. The system comprises an integrated unit, a recognition pen, a Memory card, supporting books with identification labels, and system software. The product aims to improve the English and Chinese language teaching quality in both primary and middle schools; 2) intelligent speech evaluation system. Based on speech recognition, natural language understanding, and multiple disciplines including linguistics, testing, and statistics, the system can analyse the speech characteristics of a user and score the pronunciation of the user. In addition, the system can point out the errors and flaws in the user's speech and give suggestions.

Currently, iFlytek has provided B2B education solution to 7,000 schools, compared to 3,000 schools in 2015. iFlytek also provides B2C services through supplying equipment for educational institution and charging add-on service fee. Notably, we think the potential margin upside might come from education business as its smart class and smart education platform might start monetarizing in the near future.

3 Internet & Telecom (13% of total sales in FY15)

iFlytek provides Internet & Telecom products and value-added services to China telecom operators such as China Mobile and China Unicom. Telcos would supply

Exhibit 22: Voice-based software revenues and y-y growth Exhibit 23: Voice-based software revenue contribution, GM


Exhibit 24: Education software revenues and y-y growth Exhibit 25: Education revenues contribution and GM


0

10

20

30

40

50

60

70

80

0

200

400

600

800

1,000

1,200

2012 2013 2014 2015 2016E 2017E 2018E

(y-y %)(RMB m) Revenues (LHS) Growth (RHS)

50

60

70

80

90

0

2

4

6

8

10

12

14

16

18

2012 2013 2014 2015 2016E 2017E 2018E

(%)(%) Contribution (LHS) GM (RHS)

0

20

40

60

80

100

120

140

160

180

200

0

500

1,000

1,500

2,000

2,500

3,000

2012 2013 2014 2015 2016E 2017E 2018E


0

10

20

30

40

50

60

70

0

5

10

15

20

25

30

35

40

2012 2013 2014 2015 2016E 2017E 2018E


55



intelligent voice applications and solutions to its end users, supported by iFlytek’s technology. Revenues of internet & telecom mainly come from profit-sharing with telecom operators’ service revenues.

4 Public & Government (10% of total sales in FY15)

iFlytek’s applications in public & government areas are public security and smart city information management. In public security area, iFlytek has signed strategic collaboration agreement with several municipal public safety bureaus to provide recognition technology for public safety solutions. Also, iFlytek supplies smart city information management platforms in several provinces, including Anhui, Jinlin, and Jiangxi province.

5 System Integration (30% of total sales in FY15)

iFlytek’s system integration business line mainly includes hardware products such as information engineering and audio & video monitoring. iFlytek acts as system integrator and receives order by project base from central or local government in differenct areas, such as hospital, subway station, airport, and so on. It is notably that system integration business has the lowest gross margin performance of less than 20% among the company’s overall portfolio. As the company will focus more on voice-based business, we believe system integration contribution will decline in the following years, which improves iFlytek’s margin performance.

Exhibit 26: Internet & Telecom revenues and y-y growth Exhibit 27: Internet & Telecom revenues contribution and GM


Exhibit 28: Public & Government revenues and y-y growth

Exhibit 29: Public & Government revenues contribution and GM


0

10

20

30

40

50

60

0

200

400

600

800

1,000

1,200

2012 2013 2014 2015 2016E 2017E 2018E


79

80

81

82

83

84

85

86

87

0

2

4

6

8

10

12

14

16

2012 2013 2014 2015 2016E 2017E 2018E


(40)

(20)

0

20

40

60

80

100

120

0

100

200

300

400

500

600

700

800

900

2012 2013 2014 2015 2016E 2017E 2018E


72

74

76

78

80

82

84

86

0

2

4

6

8

10

12

14

2012 2013 2014 2015 2016E 2017E 2018E


56



6 Big data service (5% of total sales in FY15)

iFlytek introduced its new business big data service, which includes communication data analysis and big data advertising platform since 2015. For communication data analysis service, it is a contract-based service, which includes the content and length of the service. For big data advertising platform, iFlytek charges its clients based on the frequency of advertisement showed or advertisement clicked by end-users. We forecast the big data service to remain stable in the following years.

Exhibit 30: System Integration revenues and y-y growth Exhibit 31: System Integration revenues contribution and GM


Exhibit 32: Big data service revenues and y-y growth Exhibit 33: Big data service revenues contribution and GM


0

10

20

30

40

50

60

70

0

200

400

600

800

1,000

1,200

1,400

1,600

2012 2013 2014 2015 2016E 2017E 2018E


0

5

10

15

20

25

30

0

5

10

15

20

25

30

35

40

2012 2013 2014 2015 2016E 2017E 2018E


0

10

20

30

40

50

60

0

50

100

150

200

250

300

2015 2016E 2017E 2018E


0

5

10

15

20

25

30

35

40

0

1

2

3

4

5

6

2015 2016E 2017E 2018E


57



Competitive landscape

We see major competition from both Nuance as well as Baidu in China as the rest of local players have small base and cannot be a big threat to iFlytek, in our view.

Competition with Nuance

Founded in 1995, Nuance also provides voice recognition solutions and natural language understanding technologies. It offers multiple solutions, including automated speech recognition, dialog and information management, biometric speaker authentication, text-to-speech, and optical character recognition (OCR). According to SIAC, Nuance is the global leading voice recognition solution supplier with largest market share of 32% in 2015.

Strategy and business model slightly different We think the key difference between iFlytek and Nuance is the business model. Nuance has a broader exposure to different regions with different language while iFlytek only focus on Chinese language in China. Meanwhile, iFlytek applied its speech recognition know-how into few industries but Nuance has more exposure to auto market.

Nuance is less profitable Despite a higher gross margin performance compared to iFlytek, Nuance’s opex ratio came in at 45%+ over 2011-2015, compared to iFlytek’s 36-40% level. From the financial numbers, we think the higher opex ratio of Nuance is mainly due to higher SG&A expense at 30%+. Furthermore, Nuance is suffering from high interest expense, which leads to loss-making bottom-line during the past three years.

We sees a strong revenues growth with 46% CAGR over 2012-2015, compared to Nuance’s 11% in the same period, thanks to fast development of voice recognition in China market. In China market, Nuance only has 3% market share in 2015, while iFlytek is the biggest player with more than 40% market share. We believe iFlytek will continue to benefit from market expansion and remain as the leading player in China.

Chinese means higher entry barrier for overseas player We think iFlytek still has advantage on China local language on (1) big Chinese data base (2) better understanding of different local accents, and better relationship with local customers. In China, the speech recognition market is dominated by iFlytek and Baidu with 72% market share, followed by Apple 7%, Nuance 3%, Xiaoi 小 i 机器人 3%, Jietong Huasheng 捷通华声 2%, Haitian Ruisheng 海天瑞声 2%, Zongke Xinli 中科信利 1%, and others 10%.

Exhibit 34: Revenues and y-y growth comparison Exhibit 35: Nuance has a higher SG&A expense ratio

Source: Companies Source: Companies

0

10

20

30

40

50

60

70

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

2012 2013 2014 2015

(y-y %)(RMB m)iFlytek sales (LHS)Nuance sales (LHS)iFlytek growth (RHS)Nuance growth (RHS)

0

5

10

15

20

25

30

35

40

2011 2012 2013 2014 2015

(%) SG&A (iFlytek) SG&A (Nuance)R&D (iFlytek) R&D (Nuance)

58



Baidu is a potential threat

In China, Baidu is a strong competitor as it has demonstrated its commitment in its Baidu World 2016 (百度世界大会 2016) on artificial intelligence (AI) through continued development of its Baidu Brain. Baidu Brain is a super large scale neural network with massive computing power and vast access to Chinese language data, and is now mature enough to begin commercialization. We noted voice recognition is one of its focusing areas, together with image recognition, natural language processing, and user profiling. Recently, Baidu plans to launch its voice mobile keyboard app on Android in Nov this year and has collaborated with Harman on smart sound speakers.

In terms of voice recognition quality, especially under noisy environment, we think iFlytek has better performance in auto market with total auto model reaching 30 models already by the end of 2015. iFlytek targets to penetrate into 100 auto model in 2016, on the back of its superior quality and accuracy. On the other hand, we have not seen clear breakthrough from Baidu in auto OEMs yet. Our view is that Baidu can have better performance on picture recognition or facial recognition but is still behind iFlytek in voice area.

Exhibit 36: Margins comparison Exhibit 37: Nuance revenues mix in the past 5 years

Source: Companies Source: Nuance

Exhibit 38: Voice recognition for automotive

Company Voice Biometrics Available Virtual Assistant Available Business Areas OEMs Working Tier-I Suppliers

Nuance Prototype available Available for web applications Automotive, healthcare, wearables, and connected living

All major OEMs All major Tier-I suppliers

iFlytek Under development Under development Automotive and Smartphones China OEMs Not available

Sensory Yes Under development Wearables, automotive vehicles, consumer electronics, and banking

In talks with mass-market OEMs

Sony, Samsung, TomTom, LG, and Motorola

VoiceBox Under development and expected by 2018

Under development Automotive, smartphones, connected homes, and wearables

Fiat, Mazda, Renault, Toyota, and Chrysler

TomTom, Pioneer, AT&T, Elektrobit, and Fujitsu Ten

iNago No Yes, for aftermarket navigation Automotive, smartphones, wearables, and consumer electronics

Toyota Clarion, Samsung, Fujitsu Ten, NNG, Toyota, and Kenwood

Sources: Frost & Sullivan; BNP Paribas estimates

(10)

0

10

20

30

40

50

60

70

2011 2012 2013 2014 2015

(%) iFlytek GM iFlytek NMNuance GM Nuance NM

0

10

20

30

40

50

60

70

80

90

100

2011 2012 2013 2014 2015

(%) Healthcare Mobile and Consumer Enterprise Imaging

59



Initiate at HOLD with a target price of RMB31

We initiate coverage of iFlytek with a HOLD rating and TP of RMB31, implying 9.0% potential upside from current levels. We expect the autonomous driving technology to drive increasing demand for voice recognition for the foreseeable future. We believe iFlytek, as the largest voice recognition provider in China, should maintain its leading position to benefit from broader application in different market. However, it is trading at 52x FY17 PE, broadly in line with peers’ 54x average, and we see its risk-reward profile is less convincing at this moment. Our TP of RMB31 is based on 57x FY17 P/E, in line with historical average since 2011 as we expect its FY16-18 earnings CAGR of 34%, which is in line with historical earnings CAGR in 2010-15. We suggest investors to wait for a better entry point.

Exhibit 39: 12m forward P/E band Exhibit 40: 12m forward P/BV band


Exhibit 41: Historical 12m forward P/E band Exhibit 42: iFlytek stock performance versus index


0

10

20

30

40

50

60

Sep-12 Sep-13 Sep-14 Sep-15 Sep-16

(RMB)

45x

57x

73x

155x

0

10

20

30

40

50

60

Sep-12 Sep-13 Sep-14 Sep-15 Sep-16

(RMB)

3.0x

5.5x

6.5x

11.3x

30

50

70

90

110

130

150

Sep-13 Sep-14 Sep-15 Sep-16

(x)

-1SD

Avg.

+1SD

+2SD

+3SD

050

100150200250300350400450500

Jan-

11Ap

r-11

Jul-1

1O

ct-1

1Ja

n-12

Apr-1

2Ju

l-12

Oct

-12

Jan-

13Ap

r-13

Jul-1

3O

ct-1

3Ja

n-14

Apr-1

4Ju

l-14

Oct

-14

Jan-

15Ap

r-15

Jul-1

5O

ct-1

5Ja

n-16

Apr-1

6Ju

l-16

(%) Shenzhen Index iFlytek

60



Exhibit 43: Valuation comparison Name BBG code CCY Price Mkt cap Earning growth ------- EPS ------ ------- P/E ------- ------ P/BV ------ ------ ROE ------ ---- PEG ---- - EV/EBITDA -

2016E 2017E 2016E 2017E 2016E 2017E 2016E 2017E 2016E 2017E 2016E 2017E 2016E 2017E

(LC) (USD m) (%) (%) (LC) (LC) (x) (x) (x) (x) (%) (%) (x) (x) (%) (%)

iFlytek* 002230 CH RMB 28.4 5,517 28 34 0.41 0.55 68.8 51.5 5.6 5.2 8.4 10.4 2.5 1.5 54 41

Nuance NUAN US USD 14.4 4,022 n.m. 5 1.51 1.64 9.5 8.8 2.2 2.2 20.9 23.8 n.a. 1.9 53 57

Kingsoft 3888 HK HKD 19.0 3,128 62 86 0.44 0.85 37.0 19.3 2.7 2.4 4.2 13.3 0.6 0.2 31 18

Yonyou 600588 CH RMB 22.5 4,962 30 31 0.28 0.40 81.3 56.3 5.6 5.4 7.1 8.9 2.8 1.8 18 14

Digiwin 300378 CH RMB 27.2 1,079 478 44 0.38 0.57 71.9 47.8 6.0 5.5 6.8 8.0 0.2 1.1 n.a. n.a.

Navinfo 002405 CH RMB 22.3 3,555 76 31 0.20 0.29 109.7 76.5 2.0 5.0 5.4 6.5 1.4 2.4 17 15

Thunder Software 300496 CH RMB 48.3 2,928 17 99 0.35 0.69 138.7 70.2 10.3 14.5 13.0 20.9 7.9 0.7 252 96

Wisesoft 002253 CH RMB 31.0 1,052 53 51 0.21 0.30 145.7 103.5 5.0 4.7 3.7 5.2 2.8 2.0 n.a. n.a.

Average industry peers 106 48 82.8 54.2 4.9 5.6 8.7 12.1 0.8 1.1 71 40

Hikvision* 002415 CH RMB 24.2 22,363 20 22 1.15 1.40 21.0 17.3 5.9 4.8 31.6 30.5 1.1 0.8 17 14

Luxshare* 002475 CH RMB 20.2 5,695 13 22 0.65 0.79 31.2 25.5 5.7 4.7 20.0 20.2 2.4 1.1 22 18

Jinlong* 300032 CH RMB 17.3 2,110 (5) 5 0.38 0.40 45.3 43.2 5.9 5.3 13.7 12.9 n.m. 8.7 32 31

Joyson* 600699 CH RMB 33.7 3,476 73 63 0.71 1.15 47.7 29.2 2.6 2.4 8.3 8.4 0.7 0.5 22 18

GoerTek* 002241 CH RMB 30.1 6,933 18 21 0.97 1.17 31.1 25.8 3.8 3.0 13.7 12.9 1.7 1.2 19 16

Desay* 000049 CH RMB 42.0 1,288 23 10 1.38 1.52 30.4 27.7 7.2 5.8 26.6 23.1 1.3 2.9 18 16

Inspur* 000977 CH RMB 26.0 3,922 38 9 0.62 0.67 41.8 38.5 7.6 6.5 19.9 18.1 1.1 4.5 25 23

Sunwoda 300207 CH RMB 15.9 3,022 60 50 0.42 0.61 38.2 25.9 8.5 6.6 22.5 26.4 0.6 0.5 22 16

Sunway 300136 CH RMB 26.0 3,718 132 46 0.56 0.83 46.9 31.3 12.7 9.2 27.7 30.3 0.4 0.7 34 23

Sunlord 002138 CH RMB 18.6 2,060 43 31 0.51 0.66 36.6 28.0 5.2 4.5 14.2 16.3 0.9 0.9 23 21

Everwin 300115 CH RMB 27.8 3,780 52 38 0.84 1.16 33.0 24.0 6.3 5.0 17.8 20.5 0.6 0.6 17 12

O-Film 002456 CH RMB 36.1 5,733 89 45 0.87 1.24 41.4 29.2 5.5 4.7 13.6 17.2 0.5 0.7 21 16

Average China peers 46 30 37.1 28.8 6.4 5.2 19.1 19.8 0.8 1.0 23 19

Priced as at 29 Sep 2016 Sources: Bloomberg consensus; *BNP Paribas estimates

61



DCF analysis

We crosscheck our target price with a DCF valuation, which gives a value of RMB31/share, based on a WACC of 7.4%.

Exhibit 44: Free cash flow analysis

2016E 2017E 2018E 2019E 2020E 2021E 2022E 2023E 2024E 2025E Terminal val.

Revenue 3,524 4,944 6,906 9,669 13,536 18,274 23,756 30,883 37,060 40,025 EBITDA 502 751 1,090 1,450 1,895 2,558 3,088 4,015 4,447 4,703 Less: Changes in working capital 422 378 526 197 1,006 1,232 1,425 1,853 1,606 771 Less: Capex 600 500 400 503 650 786 903 926 926 800 Less: Tax 62 81 114 80 104 144 173 227 251 264 FCF (583) (207) 50 670 136 397 587 1,009 1,664 2,868 3,018

Discount factor (%) 0.96 0.90 0.84 0.78 0.72 0.67 0.63 0.58 0.54 0.51 0.49

PV of FCF (375) (45) 167 638 207 369 462 676 985 1,526 34,512


Exhibit 45: WACC assumption of 7.4% is in line with Bloomberg’s WACC calculation Target capital structure (%) Debt to total capitalization 2

Equity to total capitalization 98

Debt to Equity ratio 2

Cost of equity (%) Risk-free rate 2.9

Market risk premium 6.1

Levered Beta (x) 0.76

Cost of equity (%) 7.6

Cost of Debt (%) Cost of debt 1.9

Taxes 6.0

After tax cost of debt (%) 1.8

WACC (%) 7.4


Exhibit 46: DCF implied value: RMB31/share

(RMB m)

Enterprise value Present value of free cash flow 3,085

Terminal value 70,765

Discount factor 0.49

Present value of terminal value 34,512

% of Enterprise value (%) 92%

Enterprise value 37,597

Less: Total debt 37

Plus: Cash and cash equi. 2,645

Net debt (2,607)

Implied equity value 40,204

Outstanding shares 1,285

Implied share price 31


62



Financial Analysis

1H16 review and FY16 outlook

The company reported 1H16 earnings of RMB256m (EPS of RMB0.19), 79% y-y growth. 1H16 revenue went up by 41% y-y to reach RMB1,462m. It is notably that selling expense went up by 65.3% y-y in 1H16, versus 40.7% y-y sales growth in the same period, which was mainly due to higher employee salary and other admin expense. We currently model FY16 revenue to reach RMB3,524m, up 41% y-y with gross margin from 48.9% in 2015 to 48.7% in 2016. We believe 2016 remains a growing year with 2H16 revenue to increase by 41% h-h to reach RMB2,062m, driven by continued expansion in China voice recognition market.

Profitability and margin

We model a slightly gross margin increase into 2016-18 due to more revenues contribution from its voice-based software and education-related product, which bring in a better margin performance. We expect iFlytek to maintain its operating margin at above 10% level on favourable product mix and stable opex ratio. We think iFlytek would have a fast growth and we model its net income to grow by 28%, 34%, and 40%, respectively in 2016-18.

Exhibit 47: Revenues mix by business Exhibit 48: Revenues and y-y growth


Exhibit 49: Margins Exhibit 50: Net income growth

Sources: iFlytek, BNP Paribas estimates Sources: iFlytek, BNP Paribas estimates

0

20

40

60

80

100

2013 2014 2015 2016E 2017E 2018E

(%)

Voice-based software EducationInternet & Telecom Public & GovernmentSystem Integration Big data serviceOthers

0

10

20

30

40

50

60

70

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

2013 2014 2015 2016E 2017E 2018E


0

10

20

30

40

50

60

2012 2013 2014 2015 2016E 2017E 2018E

(%) GM OPM NM

0

10

20

30

40

50

60

0

200

400

600

800

1,000

1,200

2012 2013 2014 2015E 2016E 2017E 2018E

(y-y %)(RMB m) Net income (LHS) Growth (RHS)

63



Exhibit 51: Quarterly P&L

1Q15 2Q15 3Q15 4Q15 1Q16 2Q16 3Q16E 4Q16E 1Q17E 2Q17E 3Q17E 4Q17E 2015 2016E 2017E 2018E

(RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m) (RMB m)

Sales 406 633 635 827 633 829 920 1,141 922 1,150 1,297 1,575 2,501 3,524 4,944 6,906

Gross profit 230 283 299 411 301 397 449 569 447 569 642 791 1,223 1,717 2,449 3,445

Op. profit 32 59 55 136 41 92 110 170 92 149 171 239 282 412 651 980

Pre-tax profit 77 86 102 199 78 207 133 191 132 189 211 279 465 609 811 1,140

Net income 67 77 81 202 71 185 112 176 108 169 186 264 425 544 727 1,016

EPS 0.05 0.06 0.06 0.16 0.05 0.14 0.08 0.13 0.08 0.13 0.14 0.20 0.33 0.41 0.55 0.77

Margins (%) Gross margin 56.6 44.7 47.0 49.7 47.6 47.9 48.8 49.9 48.5 49.4 49.5 50.2 48.9 48.7 49.5 49.9

Op. margin 8.0 9.3 8.6 16.4 6.5 11.0 11.9 14.9 10.0 12.9 13.2 15.2 11.3 11.7 13.2 14.2

Pre-tax margin 19.0 13.6 16.1 24.1 12.4 24.9 14.4 16.7 14.4 16.4 16.3 17.7 18.6 17.3 16.4 16.5

Net margin 16.4 12.1 12.7 24.4 11.2 22.4 12.1 15.5 11.7 14.7 14.3 16.8 17.0 15.4 14.7 14.7

Change (q-q %) Sales (39) 56 0 30 (23) 31 11 24 (19) 25 13 21 Gross profit (44) 23 6 38 (27) 32 13 27 (21) 27 13 23 Op. profit (81) 82 (7) 148 (70) 123 20 55 (46) 61 15 40 Pre-tax profit (61) 11 18 95 (61) 164 (36) 44 (31) 43 12 32

Net income (63) 15 5 150 (65) 162 (40) 58 (39) 57 10 42 Change (y-y %) Sales 55 64 39 23 56 31 45 38 46 39 41 38 41 41 40 40

Gross profit 46 44 35 (1) 31 40 50 38 48 43 43 39 24 40 43 41

Op. profit 12 49 4 (21) 27 56 100 25 125 63 56 41 (3) 46 58 51

Pre-tax profit 25 19 0 1 1 140 30 (4) 69 (9) 59 46 7 31 33 41

Net income 31 16 (1) 11 6 142 39 (13) 53 (9) 67 50 12 28 34 40


Healthy balance sheet and cash flow

We expect iFlytek to maintain a healthy cash flow and balance sheet, driven by stable growth of voice recognition business. It has maintained net cash position in the past four years; we think it will continue to maintain at a healthy balance sheet position on better earnings momentum. Meanwhile, we estimate a stable capex in 2016-18 as iFlytek would prepare its AI development in the following year.

Exhibit 52: Earnings comparison

------------------------- 2015E ------------------------- ------------------------- 2016E ------------------------- ------------------------- 2017E -------------------------

BNP Cons Diff. BNP Cons Diff. BNP Cons Diff.


Sales 3,524 3,501 1 4,944 4,919 1 6,906 6,863 1

Gross profit 1,717 1,731 (1) 2,449 2,432 1 3,445 3,423 1

Operating profit 412 432 (5) 651 613 6 980 853 15

Pre-tax profit 609 591 3 811 789 3 1,140 1,054 8

Net profit 544 532 2 727 710 2 1,016 948 7

Margins (%)

Gross margin 48.7 49.4 49.5 49.4 49.9 49.9 Operating margin 11.7 12.3 13.2 12.5 14.2 12.4 Pre-tax margin 17.3 16.9 16.4 16.0 16.5 15.4

Net margin 15.4 15.2 14.7 14.4 14.7 13.8

Sources: Bloomberg consensus; BNP Paribas estimates

64



Exhibit 53: Net debt and net-debt-to-equity ratio


Exhibit 54: Capex Exhibit 55: FCF performance


(60)

(50)

(40)

(30)

(20)

(10)

0

(3,000)

(2,500)

(2,000)

(1,500)

(1,000)

(500)

0

2012 2013 2014 2015 2016E 2017E 2018E

(%)(RMB m) Net debt (LHS) Net debt to equity (RHS)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0

100

200

300

400

500

600

700

800

900

1,000

2012 2013 2014 2015 2016E 2017E 2018E

(%)(RMB m) Capex (LHS) Capex to sales (RHS)

(0.4)(0.3)(0.3)(0.2)(0.2)(0.1)(0.1)0.00.10.10.20.2

(500)

(400)

(300)

(200)

(100)

0

100

200

300

2012 2013 2014 2015 2016E 2017E 2018E

(RMB)(RMB m) Free cash flow (LHS)Free cash flow per share (RHS)

65



Sensitivity analysis

Sales and gross margin - We also see potential earnings downside from a lower gross margin as a result of unfavourable product mix. In 2016, an additional 1ppt on gross margin could lead to 5.6-6.0% earnings downside. Meanwhile, every 1ppt increase in annual growth would positively impact earnings by 2%

Exhibit 56: Earnings sensitivity analysis on revenue growth and gross margin (FY16E)

Earnings ---------------- Annual revenue growth ---------------- Change (%) ---------------- Annual revenue growth ----------------

(RMB m) 31% 36% 41% 46% 51% (%) 31% 36% 41% 46% 51%

GM

45.7% 347 398 449 501 552

GM

45.7% (36) (27) (17) (8) 1

46.7% 376 428 481 533 586 46.7% (31) (21) (12) (2) 8

47.7% 405 459 513 566 620 47.7% (25) (16) (6) 4 14

48.7% 435 489 544 599 654 48.7% (20) (10) 0 10 20

49.7% 464 520 576 632 687 49.7% (15) (4) 6 16 26

50.7% 494 550 607 664 721 50.7% (9) 1 12 22 33

51.7% 523 581 639 697 755 51.7% (4) 7 17 28 39


Exhibit 57: Earnings sensitivity analysis on revenue growth and gross margin (FY17E)

Earnings ---------------- Annual revenue growth ---------------- Change (%) ---------------- Annual revenue growth ----------------

(RMB m) 30% 35% 40% 45% 50% (%) 30% 35% 40% 45% 50%

GM

46.5% 446 520 594 668 741

GM

46.5% (39) (29) (18) (8) 2

47.5% 487 563 638 714 789 47.5% (33) (23) (12) (2) 8

48.5% 529 606 683 760 837 48.5% (27) (17) (6) 4 15

49.5% 570 649 727 806 884 49.5% (22) (11) 0 11 22

50.5% 612 692 772 852 932 50.5% (16) (5) 6 17 28

51.5% 653 735 816 898 980 51.5% (10) 1 12 23 35

52.5% 694 777 861 944 1,027 52.5% (5) 7 18 30 41


66



DuPont analysis

Among the peers, Nuance is the direct competitors to iFlytek in the voice recognition market. Despite a loss-making performance in the past three years, the streets (based on Bloomberg) holds an optimistic view on Nuance with ROE at 20%+ level into 2016-18E. However, we note that the financial leverage ratio is higher for Nuance at over 2.5x, compared to iFlytek’s 1.3x level. We believe iFlytek has a healthier financial structure versus its peer. Moreover, iFlytek delivers a outstanding ROE performance than other China peers, mainly thanks to its better net margin.

Exhibit 58: DuPont analysis on peers

Company BBG code ----------- Net margin ----------- --------- Asset turnover --------- --------- Leverage ratio --------- ---------------- ROE ----------------

2016E 2017E 2018E 2016E 2017E 2018E 2016E 2017E 2018E 2016E 2017E 2018E

(%) (%) (%) (x) (x) (x) (x) (x) (x) (%) (%) (%)

iFlytek* 002230 CH 15.4 14.7 14.7 0.4 0.5 0.7 1.3 1.3 1.3 8.4 10.4 13.3

Nuance NUAN US 24.8 23.3 24.1 0.3 0.3 0.4 2.5 3.1 2.9 20.9 24.6 24.4

Navinfo 002405 CH 10.6 10.8 11.3 0.4 0.4 0.5 1.4 1.4 1.4 5.4 6.5 7.6

Joyson* 600699 CH 3.6 4.2 4.4 1.5 1.1 1.2 1.5 1.8 1.8 8.3 8.4 9.5

Neusoft 600718 CH 9.9 6.9 7.5 0.7 0.6 0.5 1.5 2.2 2.1 9.9 8.6 8.5

Protruly 600074 CH 27.2 27.7 27.5 0.1 0.2 0.2 1.1 1.2 1.2 11.6 15.9 19.9

Jingu 002488 CH 2.2 2.6 3.0 0.6 0.8 0.8 3.2 4.0 5.2 4.5 8.4 12.9

Kingsoft 3888 HK 4.3 10.4 11.1 0.4 0.6 0.8 3.1 2.0 1.9 4.7 13.1 16.4

Yonyou 600588 CH 8.6 10.0 11.4 0.5 0.5 0.5 1.7 1.8 2.0 7.1 8.9 11.1

Sources: Company data; Bloomberg consensus estimates, *BNP Paribas estimates

Exhibit 59: ROE comparison

Sources: Bloomberg; *BNP Paribas estimates

0

5

10

15

20

25

30

iFlytek* Nuance Navinfo Joyson* Neusoft Protruly Jingu Kingsoft Yonyou

(%) 2016E 2017E 2018E

67



Risks to our call

Downside risks

Worse-than-expected China voice recognition demand: Worse-than-expected China voice recognition demand will have negative impact on sales and earnings from voice-based application business to our current estimates.

International player to penetrate into China voice recognition market: Given a slowing global voice recognition market, Nuance might emphasize more resources to gain share in growing China voice recognition market.

Worse-than-expected growth of China smart car market: iFlytek supplies voice recognition technology for smart car’s voice control system. As such, a worse-than-expected development of smart car in China would lead to potential downside to our current estimates.

Upside risks

Better-than-expected China voice recognition demand: Better-than-expected China voice recognition demand will have positive impact on sales and earnings from voice-based application business to our current estimates.

Government policy support: If China government launches policy for local voice recognition players, it will form a natural entry barrier to international players.

Faster adoption of voice control for auto HMI. Voice control serves as a desirable method for auto HMI. A faster-than-expected adoption of voice control by auto OEMs will trigger more demand for iFlytek’s voice recognition service.

68



Company profile

Founded in 1999, iFlytek is a Chinese software enterprise focusing on speech intelligence and artificial intelligence technology. The company also provides chip products, voice messaging software, and integration service for e-government systems. The company mainly provides its service for telecommunication, financial institution, government, and other industries. iFlytek went listed on Shenzhen Stock Exchange in 2008.

Exhibit 60: Key milestone – iFlytek

Date Key event

1999 iFlytek was founded by a group of college students

2001 Number of developers reached over 100 on iFlytek intelligent speech platform

2005 Set up iFlytek research institution

2006 Won first prize at Blizzard Challenge from 2006 to 2011

2007 iFlytek Co. was officially established, and its intelligent speech chips sales reached 200k units

2008 iFlytek went listed on Shenzhen Stock Exchange

2009 Listed as National Innovative Enterprise in China

2010 Launched intelligent speech interaction platform - iFlytek cloud

2012 iFlytek cloud users reached 150m in total

2013 Strategic alliance with telco operators including China Mobile, China Telecom, and China Unicom

Source: iFlytek

Exhibit 61: Top 10 shareholders

Name Holding (%)

China mobile 13.9

Qingfeng Liu 8.1

USTC Holdings 4.3

Renhua Wang 1.9

China Investment Corporation 1.4

China Asset Management 1.1

Anhui Railway Construction 1.1

Hongwei Hu 1.0

Tao Chen 1.0

E fund management 0.9

Others 64.6

Source: iFlytek

Exhibit 62: Management profile

Current position Name Major experience Educational Background

Chairman Mr. Liu Qingfeng 刘庆峰 Chairman and founder of iFlytek. Representative of the 12th National People's congress in China.

PHD in engineering

Director Mr. Liu Xin 刘昕 Director of iFlytek and chairman of MIGU. Past experience as solution manager in IBM and department manager in China Mobile.

Master degree

Director Mr. Su Jun 苏俊 Director of iFlytek and CEO of Teamtech information system. Past experience as lecturer in University of Science and Technology of China.

Bachelor in science

Director Mr. Chen Tao 陈涛 Director of iFlytek. Executive director in iFlyteks' subsidiaries. Master in computer science

Director Mr. Wu Xiaoru 吴晓如 Director of iFlytek. Executive director in iFlyteks' subsidiaries. PHD in engineering

Director Mr. Hu Yu 胡郁 Director of iFlytek. Executive director in iFlyteks' subsidiaries. PHD

Director Mr. Hu Hongwei 胡宏伟 Director of iFlytek. Executive director in iFlyteks' subsidiaries. Bachelor in engineering

Financial officer Mr. Zhang Shaobing 张少兵 Financial officer of iFlytek. Past experience as auditor in accounting firm. Bachelor

Board secretary Mr. Jiang Tao 江涛 Board secretary of iFlytek. Executive director in iFlyteks' subsidiaries. Master in computer science

Source: iFlytek

69



Exhibit 63: Employee mix by function (2015) Exhibit 64: Employee mix by education (2015)

Source: iFlytek Source: iFlytek

Technology67%

Sales & Marketing

23%

Management6%

Others4%

Master and above26%

Bachelor62%

College degree11%

Others1%

70



Corporate governance

Board structure

Number of Independent Directors (ID) 4

Percentage of IDs in the board 36.4%

ID participation/attendance at board meetings 100%

ID participation in audit/remuneration committees 100%

ID terms (years of service, re-election/replacement procedures) 3 years

Sources: Company; BNP Paribas

Audit practices

Auditor HPTJCPA

Length of service 14 years

Reporting incidents N/A

Fee track record RMB1.5m for 2015

Policy on change of Audit firm N/A


Compensation and remuneration

Directors' remuneration vs. earnings/ROE/share performance 1.9% of 2015 earnings

Changes/stability in senior management Stable

Incidents of termination of senior management N/A

Track record on Insider sales N/A


Shareholders' rights

Communication - shareholder participation in AGMs/EGMs 34.2%

Related party transactions N/A

Voting issues - policies, incidents of rejected proposals None


71



Financial statements Iflytek

Sources: Iflytek; BNP Paribas estimates

Profit and Loss (RMB m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E

Revenue 1,775 2,501 3,524 4,944 6,906

Cost of sales ex depreciation (728) (1,192) (1,717) (2,395) (3,351)

Gross profit ex depreciation 1,048 1,309 1,807 2,549 3,555Other operating income 0 0 0 0 0

Operating costs (696) (941) (1,305) (1,798) (2,465)

Operating EBITDA 352 368 502 751 1,090Depreciation (60) (86) (90) (100) (110)

Goodwill amortisation 0 0 0 0 0

Operating EBIT 292 282 412 651 980Net financing costs 42 38 40 40 40

Associates 13 39 134 40 40

Recurring non operating income 100 145 156 120 120

Non recurring items 0 0 0 0 0

Profit before tax 434 465 609 811 1,140Tax (45) (28) (62) (81) (114)

Profit after tax 388 437 546 730 1,026Minority interests (9) (11) (2) (3) (10)

Preferred dividends 0 0 0 0 0

Other items 0 0 0 0 0

Reported net profit 379 425 544 727 1,016Non recurring items & goodwill (net) 0 0 0 0 0

Recurring net profit 379 425 544 727 1,016

Per share (RMB)

Recurring EPS * 0.32 0.33 0.41 0.55 0.77

Reported EPS 0.32 0.33 0.41 0.55 0.77

DPS 0.11 0.10 0.12 0.17 0.23

Growth

Revenue (%) 41.6 40.9 40.9 40.3 39.7

Operating EBITDA (%) 32.2 4.5 36.4 49.6 45.1

Operating EBIT (%) 29.3 (3.5) 46.1 58.0 50.5

Recurring EPS (%) 35.1 4.8 24.8 33.7 39.7

Reported EPS (%) 35.1 4.8 24.8 33.7 39.7

Operating performance

Gross margin inc depreciation (%) 55.6 48.9 48.7 49.5 49.9

Operating EBITDA margin (%) 19.8 14.7 14.2 15.2 15.8

Operating EBIT margin (%) 16.5 11.3 11.7 13.2 14.2

Net margin (%) 21.4 17.0 15.4 14.7 14.7

Effective tax rate (%) 0.0 0.0 0.0 0.0 0.0

Dividend payout on recurring profit (%) 35.0 31.0 30.0 30.0 30.0

Interest cover (x) n/a n/a n/a n/a n/a

Inventory days 75.8 75.5 72.7 67.4 66.7

Debtor days 199.0 193.6 170.0 147.7 127.3

Creditor days 248.1 210.3 165.9 128.1 105.2

Operating ROIC (%) 30.0 18.8 18.2 20.9 25.0

ROIC (%) 16.0 11.5 11.8 13.6 17.0

ROE (%) 10.8 8.5 8.4 10.4 13.3

ROA (%) 7.3 5.9 5.9 7.6 10.0*Pre exceptional pre-goodwill and fully diluted

Revenue By Division (RMB m) 2014A 2015A 2016E 2017E 2018E

Voice-based software 246 303 414 652 991

Education 662 761 1,066 1,582 2,419

Internet & Telecom 206 313 486 716 1,006

Public & Government 152 248 395 572 802

System Integration 470 757 978 1,200 1,423

Big data service 0 113 175 215 258

Others 40 7 9 7 7

72



Financial statements Iflytek

Sources: Iflytek; BNP Paribas estimates

Cash Flow (RMB m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E

Recurring net profit 379 425 544 727 1,016

Depreciation 60 86 90 100 110

Associates & minorities 0 0 0 0 0

Other non-cash items 296 213 0 0 0

Recurring cash flow 735 725 634 827 1,126Change in working capital (315) (209) (422) (378) (526)

Capex - maintenance 0 0 0 0 0

Capex - new investment (580) (890) (600) (500) (400)

Free cash flow to equity (160) (374) (388) (50) 200Net acquisitions & disposals 0 0 0 0 0

Dividends paid (95) (133) (132) (163) (218)

Non recurring cash flows (538) (467) 0 0 0

Net cash flow (793) (974) (520) (214) (18)Equity finance 124 2,307 0 0 0

Debt finance (19) 245 (50) (50) (50)

Movement in cash (688) 1,578 (570) (264) (68)

Per share (RMB)

Recurring cash flow per share 0.61 0.56 0.48 0.63 0.85

FCF to equity per share (0.13) (0.29) (0.29) (0.04) 0.15

Balance Sheet (RMB m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E

Working capital assets 1,498 2,122 2,568 3,042 3,722

Working capital liabilities (1,067) (1,379) (1,402) (1,499) (1,653)

Net working capital 431 743 1,165 1,543 2,069Tangible fixed assets 767 1,056 1,566 1,966 2,256

Operating invested capital 1,197 1,799 2,732 3,509 4,325Goodwill 608 650 650 650 650

Other intangible assets 0 0 0 0 0

Investments 167 - 513 513 513

Other assets 1,064 1,404 1,404 1,404 1,404

Invested capital 3,036 4,367 5,299 6,077 6,893Cash & equivalents (1,067) (2,645) (2,075) (1,811) (1,743)

Short term debt 9 222 172 122 72

Long term debt * 0 37 37 37 37

Net debt (1,058) (2,385) (1,865) (1,652) (1,634)Deferred tax 0 0 0 0 0

Other liabilities 193 229 229 229 229

Total equity 3,707 6,268 6,680 7,244 8,042

Minority interests 195 255 255 255 255

Invested capital 3,036 4,367 5,299 6,077 6,893

Per share (RMB)

Book value per share 3.08 4.88 5.07 5.50 6.10

Tangible book value per share 2.58 4.37 4.58 5.00 5.61

Financial strength

Net debt/equity (%) (27.1) (36.6) (26.9) (22.0) (19.7)

Net debt/total assets (%) (20.5) (28.4) (21.3) (17.6) (15.9)

Current ratio (x) 2.4 3.0 2.9 3.0 3.2

CF interest cover (x) n/a n/a n/a n/a n/a

* includes convertables and preferred stock which is being treated as debt

Valuation 2014A 2015A 2016E 2017E 2018E

Recurring P/E (x) * 90.1 85.9 68.8 51.5 36.9

Recurring P/E @ target price (x) * 98.2 93.7 75.1 56.2 40.2

Reported P/E (x) 90.1 85.9 68.8 51.5 36.9

Dividend yield (%) 0.4 0.4 0.4 0.6 0.8

P/CF (x) 46.5 50.4 59.1 45.3 33.3

P/FCF (x) (213.5) (97.6) (96.5) (741.9) 187.0

Price/book (x) 9.2 5.8 5.6 5.2 4.7

Price/tangible book (x) 11.0 6.5 6.2 5.7 5.1

EV/EBITDA (x) 94.6 93.5 71.4 48.0 33.1

EV/EBITDA @ target price (x) 103.4 102.5 78.2 52.5 36.2

EV/invested capital (x) 11.0 7.9 6.8 5.9 5.2* Pre exceptional & pre-goodwill and fully diluted

73

Bai du BIDU U S BNP PARIBAS Vey- Ser n Ling

30 SEPTEMBER 2016 CHINA / INTERNET

BAIDU BIDU US

New technologies = new hope Click here to see Interactive Scenarios

KEY STOCK DATA


Well-positioned to benefit from autonomous vehiclesOur China tech analyst Jimmy Huang sees five key components inautonomous vehicles (AV): 1) sensors/radars; 2) deep learning (AI);3) high-definition mapping; 4) human machine interfaces (HMI); and5) motion control (see Autonomous vehicles: a megatrend in themaking, 30 Sep 2016). We think Baidu has leading technologies in three of the five: deep learning (AI), high-definition mapping and HMI.

Initial commercialization in 2018 then mass production in 2021Baidu has an incremental AV plan, with public-use ‘conditional’autonomous vehicles on fixed routes/areas in China’s select cities by2018, and mass production of driverless electric cars by 2021. Itplans to outsource vehicle production, and is also collaborating withNvidia to build an open platform for autonomous driving that deliverscloud-based 3D maps and AI capabilities to help navigation (thesystem will be available to automobile OEMs).

Abundant commercial opportunities underpinned by AIBaidu has lagged large cap Chinese internet peers in delivering newcommercial innovations, and this is reflected in relative share priceperformances. However, we think Baidu’s early investment in AImeans it is now better placed for a myriad of commercialisationopportunities as we expect AI to become a foundation technologyinfrastructure that will be transformative for many industries.

Raising our target price to USD230: BUY maintainedWe adjust our 2016E/17E/18E EPS estimates by -10%/+1%/+10%and raise our target price by 18% to USD230 based on DCF analysis (3% terminal growth and 10.4% WACC unchanged). Excluding netcash and stakes in Ctrip and iQiyi, we estimate Baidu’s core searchbusiness trades at 11.3x 2017E P/E, which we find attractive given its27% y-y revenue growth in 2015. Also, we believe Baidu’s leadingposition in AI technology will lead to significant financial upside.

See our interactive scenarios for Baidu here

Vey-Sern Ling [email protected] +852 2825 1802

BUY UNCHANGED

TARGET PRICE USD230.00

CLOSE USD182.23

UP/DOWNSIDE +26.2%

PRIOR TP USD195.00

CHANGE IN TP 17.95%


TARGET PRICE (%) 21.5 POSITIVE 18

EPS 2016 (%) (8.3) NEUTRAL 9

EPS 2017 (%) 11.0 NEGATIVE 4

2015A 2016E 2017E 2018E

Revenue 66,382 72,385 93,685 114,737

Rec. net profit 33,664 9,807 17,416 25,862

Recurring EPS (RMB) 96.40 28.31 50.28 74.66

Prior rec. EPS (RMB) 96.40 31.40 49.99 68.12

Chg. In EPS est. (%) 0.0 (9.8) 0.6 9.6

EPS growth (%) 156.3 (70.6) 77.6 48.5

Recurring P/E (x) 12.6 42.9 24.2 16.3

Dividend yield (%) 0.0 0.0 0.0 0.0

EV/EBITDA (x) 22.6 21.9 14.5 9.3

Price/book (x) 5.3 4.6 3.8 3.0

Net debt/Equity (%) (39.2) (43.5) (52.3) (60.6)

ROE (%) 51.1 11.4 17.2 20.7


Absolute (%) 4.9 10.8 33.3

Relative to country (%) 0.7 (5.7) 16.4

Next Results



Free float (%) 68

Major shareholder Robin Yanhong Li (16%)

12m high/low (USD) 217.97/137.41


ADR ticker -


Issued shares (m) 347

YE Dec (RMB m)

October 2016

(20)

3

25

48

70

130

153

175

198


Baidu Rel to MSCI China(USD) (%)

74

http://www.trefis.com/stock/bidu/model/bnp?easyAccessToken=PROVIDER_0d481727e52d9e96b5872c070c8e2dab0b0c8834

http://www.trefis.com/stock/bidu/model/bnp?easyAccessToken=PROVIDER_0d481727e52d9e96b5872c070c8e2dab0b0c8834

Baidu BIDU US Vey-Sern Ling


Investment thesis

Baidu is the first-mover in developing artificial intelligence (AI) technologies in China. It formed AI labs in 2013, and by 2015 had built one of the largest computing neural networks in the world (Baidu Brain) utilizing large-scale GPU parallelization.

The early investment together with Baidu's access to Chinese language search data provides it with a strong head start in the commercialization of AI. We believe AI in future will become a foundation technology infrastructure that has many applications across industries. One such transformative application is in autonomous driving, where Baidu is planning to launch 'conditional' autonomous public shuttles by 2018.

While business models are still unclear at the moment, we think the market potential for autonomous driving systems is huge, especially in China which is relatively difficult to penetrate for foreign tech companies, and where Baidu is the leading player. Our conservative illustration suggests a successful commercialisation of autonomous driving alone could increase Baidu’s revenues by over 30%.

Catalyst

Potential positive share price catalysts include better-than-expected revenue growth from increasing mobile monetisation (from both its core search business and its emerging O2O business), and margin recovery and earnings growth as investments taper towards normalised levels.

Risk to our call

Key risks to our DCF-based TP include: 1) aggressive competition in mobile search impacting Baidu's traffic share and hampering monetisation ramp-up, leading to lower-than-expected revenue growth, 2) taking longer than expected to secure mobile dominance and mobile user stickiness, leading to sustained marketing expenses and weaker margin recovery and 3) continued high level of investment in O2O initiatives without clear returns on investment.


Baidu is one of the largest web services companies in China. The company was listed on NASDAQ on 5 Aug 2005. In addition to its core web search service, the company has a broad range of products and services on both PC and mobile, including social products, app store, online video, online travel, group buy, as well as map and cloud services.

2015 2016E 2017E 2018E

Total revenue (RMB m) 66,382 72,385 93,685 114,737

Gross profit margin (%) 59 51 50 51

EBIT margin (%) 18 15 19 24

Key executives Age Joined Title

Robin Yanhong Li 47 2000 Chairman and Chief Executive Officer Jennifer Xinzhe Li 48 2008 Chief Financial Officer Ya-Qin Zhang 50 2014 President

http://www.baidu.com Sources: Baidu; BNP Paribas estimates


------ Base ------ ------- Best ------- ----- Worst -----

2016E 2017E 2016E 2017E 2016E 2017E

Total revenue (RMB m) 72,385 93,685 79,624 103,053 65,147 84,316

Change in total revenue (%) 0 0 10 10 (10) (10)

EBIT margin (%) 15.4 19.0 20.4 24.0 10.4 14.0

Change in EBIT margin (ppt) 0 0 5 5 (5) (5)

EPS (RMB) 28.31 50.28 43.03 70.31 15.68 32.96

Change (%) 52.0 39.8 (44.6) (34.5)

Source: Baidu Sources: Baidu; BNP Paribas estimates

Event calendar

Date Event

Oct-16 Google 3Q16 results

Oct-16 Baidu 3Q16 results

Oct-16 Alibaba 3Q16 results

Nov-16 Tencent 3Q16 results

Key earnings driver is total revenue and EBIT margin, in our view.

We estimate a 10% increase in total revenue and 5ppt rise in EBIT margin would raise 2016E/2017E EPS by 52.0%/39.8%, all else being equal; similarly a 10% and 5ppt decrease would reduce 2016E/2017E EPS by 44.6%/34.5%.

Online marketing service - 96.5%

Others - 3.5%

75



New technologies = new hope

Baidu lacked new commercial innovations in the past

In the past year Baidu’s share price performance has lagged its peers Tencent and Alibaba significantly. While Baidu was up close to 40% y-y as of 29 September 2016, Tencent and Alibaba were up over 60% y-y and 80% y-y respectively.

If we stretch the comparison over a longer period, Tencent’s share price has increased more than six-fold over the past five years while Baidu only managed to return close to 40%.

We think Baidu has underperformed its peers because it has not been able to generate new growth outside of its core search business.

Tencent. Through organic innovation and strategic acquisitions in the past five years, Tencent has evolved from a PC game publisher into a mobile social network with strong market positions in entertainment, O2O and finance. This enabled Tencent to deliver revenue growth acceleration through mobile games and advertising.

Alibaba. We believe investors in the past year have started to appreciate Alibaba’s value beyond its ecommerce marketplace. Alibaba has demonstrated successful organic innovation (AliCloud, Ant Financial), and made a large number of strategic investments (logistics, digital entertainment and healthcare) which are starting to look coherent and synergistic to investors.

Baidu. Baidu, on the other hand, made investments where returns seemed unclear, such as 91 Wireless, Nuomi, O2O and food delivery, and its core search business faced headwinds with new competition, mobile transition, and regulatory issues.

Leading position in artificial intelligence technologies could help Baidu

We think Baidu’s early investment in artificial intelligence (AI) technology and infrastructure, together with its unparalleled access to Chinese language search query data, provides it with a strong head start in the commercialization of AI. We believe AI could potentially be transformative for many industries and Baidu is in a better position than its domestic internet peers to benefit from countless commercialisation opportunities in the future.

First-mover advantage. Baidu is the first-mover in developing artificial intelligence technologies in China. Its Beijing Deep Learning Lab was launched in 2013. In May 2014, Baidu hired Andrew Ng as its Chief Scientist to head up its Silicon Valley AI Lab. Andrew Ng previously led the Google Brain project and is one of the world’s leading AI experts. At the same time the Beijing Big Data Lab

Exhibit 1: BAT share prices in the past year Exhibit 2: Baidu and Tencent share prices in the past 5 years

Note: As of 29 Sep 2016 Source: Bloomberg

Note: As of 29 Sep 2016 Source: Bloomberg

(20)

0

20

40

60

80

100

Sep-

15

Oct

-15

Nov

-15

Dec

-15

Jan-

16

Feb-

16

Mar

-16

Apr-1

6

May

-16

Jun-

16

Jul-1

6

Aug-

16

Sep-

16

(%) Baidu Tencent Alibaba

(100)

0

100

200

300

400

500

600

Aug-

11

Feb-

12

Aug-

12

Feb-

13

Aug-

13

Feb-

14

Aug-

14

Feb-

15

Aug-

15

Feb-

16

Aug-

16

(%)Baidu Tencent

76



was established. The three laboratories were collectively known as the Baidu Research Institute, and focused on deep learning, high performance computing, voice and image recognition, natural language processing, big data, robotics, and human-computer interaction.

Technology and capital barrier. Baidu was one of the first technology companies in the world to use large-scale GPU parallelization to build deep neural networks (Baidu Brain), even before GPU usage became the mainstream computing unit for AI. The performance of neural-network-based AI systems is directly correlated with the scale of the network and the volume of useful data fed into it. Baidu Brain built in 2015 is one of the largest computing neural networks in the world with about 100k servers, 100b neural connections, and is 100x larger than Google Brain (built in 2012).

Access to relevant data. “Supervised learning” AI algorithms require large data samples to “train” the network to produce the desired output with high accuracy. While large neural networks could be built by anyone with sufficient capital and technical know-how, the availability of useful training data is often a significant barrier to developing these systems. To this end Baidu has unparalleled access to over a billion real-world Chinese language search queries per day. Search data is particularly useful because users provide unstructured queries as input to the search engine, and click on the “answer” to their queries.

“Artificial intelligence is the new electricity”

After three years of development, Baidu showcased its AI capabilities extensively at Baidu World 2016, held in September in Beijing.

Baidu demonstrated 4 main areas of commercial application:

Voice recognition. Baidu’s Deep Speech 2 system is able to transcribe Mandarin speech with 97% accuracy in quiet environments, more accurately than a human can. This system was recognised by the MIT Technology Review as one of the top 10 breakthrough technologies in 2016. Baidu demonstrated applications such as a real-time CRM teleprompter which recognises a customer’s voice queries and provides suggested answers to a sales representative, as well as digital readers and voice transcribers.

Baidu will launch a voice-only “keyboard” IME for Android in November 2016. The app can interpret voice emotion and automatically input emoticons, as well as provide search and quick-dial when a user says a contact’s name.

Image recognition. Baidu’s Deep Image system is able to recognize faces with 99.7% accuracy. Image recognition technology can be used for digital map construction, machine vision, in driverless cars, and augmented reality applications. For example Baidu collaborated with L’Oreal on an augmented reality ad campaign where smartphone users could use the app DuSee to scan and recognise L’Oreal’s new Ultra DOUX hair care products to see interactive advertising on their smartphones. DuSee is also used in the new Mercedes-Benz E-Class marketing campaign.

Natural language processing. Baidu’s voice operated personal assistant DuEr is able to provide real-time commentary for sports events such as live basketball matches. This is done through learning from thousands of hours of NBA basketball games. DuEr is able to deliver sporting statistics and anecdotes with an artificial sense of humour. Baidu also showcased its natural language translation app (fanyi.baidu.com), which works on 27 different languages.

User profiling. Through analysing search queries as well as other traffic data on Baidu Internet properties, Baidu is able to build detailed profiles of its users. Baidu was able to apply this technology to make news recommendations to its users based on their interests. Also, more commercially, Baidu was able to help Legendary Pictures make targeted marketing efforts for the launch of the Warcraft movie in China, based on its understanding of users.

77



Commercial opportunities could be abundant in future

At the Baidu World conference, Baidu announced that it will open up the capabilities of Baidu Brain to developers (at ai.baidu.com). It will also open source its deep learning platform PaddlePaddle (PArallel Distributed Deep LEarning) to global AI researchers. This follows in the footsteps of companies like Microsoft, Google, Facebook and Amazon who have all open sourced their AI platforms in a global competition to commercialise AI.

We think AI in the future will be a foundational technology infrastructure that clearly has more potential applications than Baidu can explore on its own, hence open sourcing is a good way to attract talent and achieve wider adoption. This should eventually lead to faster commercialization of Baidu’s AI technology.

In mid-September, Baidu launched a USD200m venture capital unit called Baidu Venture to invest in early stage artificial intelligence, AR and VR projects. The unit is chaired by CEO Robin Li and run independently from Baidu’s existing investment teams, allowing it to make fast decisions.

Having said that, we believe the commercialization of AI may take many years. We think that it is possible to see incremental benefits from AI in the near term, but these would generally be in the form of enhancing user experiences of prevailing applications. For example, in the case of search engines, AI can provide better recognition of voice input queries, and generate more accurate image searches.

Truly transformative standalone AI commercial opportunities may not be upon us until several years later due to currently unsolved technical barriers. In the following section we discuss Baidu’s plan to mass produce autonomous cars by 2021, which is one of the AI opportunities that we consider to be transformative.

Autonomous cars and Baidu

The plan

Baidu plans to develop a commercially viable autonomous vehicle by 2018, and begin mass production of driverless electric cars in five years (mid-2021). The aim is to initially produce a vehicle that conforms to at least SAE level-3 standards that is capable of conditional automation. A driver’s presence is generally required in a level-3 vehicle but the car can handle most functions on its own.

Baidu's self-driving car vision is based on an incremental approach, which includes starting with small "autonomy-enabled" regions and designing autonomous vehicles to be clearly recognizable. For a start, Baidu aims to roll out public self-driving shuttle buses running a standard loop in limited areas in China by 2018. Such conditional

Exhibit 3: Level of automation of vehicle

Level Degree of automation Driver Automation system Example ADAS or Autonomous driving

0 Driver only Driver continuously in control of speed and direction.

No intervening vehicle system active N/A None

1 Assisted Driver continuously performs the longitudinal or lateral dynamic driving task.

The other driving task is performed by the system.

Park assist ADAS

2 Partial automation Driver must monitor the dynamic driving task and the driving environment at all times.

System performs longitudinal and lateral driving task in a defined use case.

Traffic Jam Assist ADAS

3 Conditional automation

Driver does not need to monitor the dynamic driving task nor the driving environment at all times; must always be in a position to resume control.

System performs longitudinal and lateral driving task in a defined use case. Recognises its performance limits and requests driver to resume the dynamic driving task with sufficient time margin.

Highway patrol Autonomous driving

4 High automation Driver is not required during defined use case.

System performs the lateral and longitudinal dynamic driving task in all situations in a defined use case.

Urban automated driving

Autonomous driving

5 Full automation No driver required. System performs the lateral and longitudinal dynamic driving task in all situations encountered during the entire journey.

Full end-to-end journey

Autonomous driving

Sources: SAE; BNP Paribas

78



automation will limit the potential for unexpected events, and routes could be expanded geographically as the system learns. The production of these shuttles will be outsourced.

For the eventual mass production of electric driverless cars Baidu will also rely on existing automakers for production. No decisions have been made yet on the manufacturers. Depending on the technological progress these cars could be SAE level-4.

Baidu is also collaborating with Nvidia to build an autonomous driving cloud-to-car open platform that will be available for branded car OEMs to use. The system will deliver cloud-based 3D maps and AI capabilities to help cars with navigation. This platform could also be rolled out to commercial vehicles such as taxis and buses.

The partners

BMW has supplied vehicles for Baidu’s testing. These are modified 3-series sedans. The partnership began in Sep 2014. BMW has not committed to any involvement in the production car in 2021.

In Aug 2016, Baidu unveiled an all-electric self-driving car based on a modified Chery EQ installed with Baidu AutoBrain software. The Chery partnership is intended for domestic launch. It is smaller and cheaper than the BMW 3-series, and taking into account government subsidies for EVs, will only cost USD9k.

In Aug 2016, Baidu and Ford announced a USD150m joint investment in Velodyne, a Silicon Valley firm that makes LIDAR sensors essential for guiding self-driving cars. LIDAR is a radar that uses light instead of radio waves. It allows for the creation of more detailed 3D maps.

At Baidu World 2016, Nvidia and Baidu announced a partnership to develop a comprehensive autonomous driving platform. Nvidia is the world’s leading designer of GPUs, which have in recent years become the key processing unit for building AI neural networks. Baidu will combine its cloud platform, 3D mapping and AI technologies with Nvidia’s self-driving vehicles platform, with the objective of making the platform open for all automobile manufacturers to use.

Nvdia revealed in mid-Sep 2016 its new Drive PX2 in-vehicle computer which will be used to power Baidu’s mapping and autonomous vehicle technology. Drive PX2 uses 10w of power and is configured from a single processor, making it one of the smallest and most efficient in-car AI computers currently.

Exhibit 4: BMW 3-series with Baidu driverless technology Exhibit 5: Chery EQ with Baidu driverless technology

Source: Baidu Source: Baidu

79



Baidu’s strengths

Our China tech analyst Jimmy Huang sees five key components in autonomous vehicles (AV): 1) sensors/radars; 2) deep learning (AI); 3) high-definition mapping; 4) human machine interfaces (HMI); and 5) motion control (see Autonomous vehicles: a megatrend in the making, 30 Sep 2016). We believe Baidu has developed leading technologies in three out of these five: deep learning (AI), mapping and HMI.

Deep learning (AI)

The AI decision-making engine is the most critical component of autonomous vehicles. This engine processes real-time sensor and map data to implement correct driving actions through actuators.

Baidu started investing in R&D of unmanned vehicles since 2014, leveraging on the foundational work it has done on AI (such as Baidu Brain). Applied to autonomous driving, this AI system is called Baidu AutoBrain. The AutoBrain software is the core of Baidu’s driverless technology, which makes use of highly automated driving maps, positioning, detection, smart decision-making and control. Intelligent decision and control helps achieve optimal path planning, make proper driving decisions according to road conditions, and maintain normal driving of vehicles.

Deep learning systems require lots of training data to increase output accuracy. For example, to teach a car to respond to road situations, many road situations will have to be fed into the neural network, with the corresponding correct driving action presented (usually by a human test driver). To this end, Baidu is planning to launch testing of its autonomous driving system in 10 cities in China (agreements already signed with Wuhu, Shanghai and Wuzhen). This will give Baidu a significant head start over competitive autonomous driving systems in training the system to local driving conditions in China.

Exhibit 6: Key components for autonomous driving vehicle

Source: BNP Paribas

Exhibit 7: Four core Baidu technologies used in Baidu AutoBrain

Source: Baidu

Intelligent decision and control

High precision map

Perception Location

80



Mapping

There are two different maps required for autonomous vehicles, and Baidu is well-positioned in both.

Baidu’s GPS navigation map (released in 2005) is already the most used mobile map in China, ahead of AutoNavi, Google and Tencent, with a market share of over 60% as of July 2016. This will generally be used by autonomous vehicles for route guidance.

Autonomous vehicles also need detailed 3D high-precision maps to understand the surrounding environment in terms of information such as traffic lights, curb height, road barriers, potholes, road elevation and so on, to enable the vehicle to make correct decisions. Baidu’s self-developed 3D map technology has achieved centimetre-level positional accuracy. In addition, the recent co-investment with Ford in Velodyne adds LIDAR capabilities. Velodyne’s equipment is capable of producing 300k-1.2m data points per second with a range up to 200m at centimetre-level accuracy.

Human machine interfaces (HMI)

HMI in vehicles is progressing quickly from physical controls to touchscreens to voice control. Voice control is clearly the most intuitive interface method that frees users from distraction, and is well suited for complex command interfaces. Baidu being the current world-leader in Mandarin speech recognition and natural language processing is clearly well positioned here.

Baidu has a suite of telematics products available today that have been implemented in the Chinese models of car brands like Mercedes Benz, BMW, Audi, VW, General Motors, Hyundai, BYD and Changan:

CarLife is an mobile app (on iOS and Android) that integrates smartphones to an in-car system (by WiFi or USB) to provide services like voice-based navigation, music streaming, phone calls and texting.

CoDriver uses Baidu’s speech recognition as a digital assistant so users can make phone calls, play music or search for locations using voice commands.

MyCar collects car data and the driver’s driving and navigation history, analyses them through the cloud, and gives drivers notifications for needed vehicle maintenance and optimizes driving routes. It also allows the driver to control the car through a private cloud, for example turning the air conditioning on remotely.

Exhibit 8: Mobile map share of times used per month Exhibit 9: Mobile map share of monthly active users

Data as of July 2016 Source: iResearch

Data as of July 2016 Source: iResearch

Tencent3%

Baidu69%

AutoNavi (Alibaba)

22%

Google6%

Tencent3%

Baidu62%

AutoNavi (Alibaba)

30%

Google5%

81



Progress so far

In Dec 2015, Baidu drove a modified BMW 3 GT autonomously through 30km of urban traffic on the fifth ring road in Beijing. The top speed during the drive was 100km/h, and the vehicle performed speed reduction, overtaking, lane switching, U-turns, and highway merging autonomously. This achieved the most complex road test conditions, the most comprehensive automatic driving actions, and the most accurate environmental understanding in China at that time.

In April 2016, Baidu formed a self-driving car team in Silicon Valley to focus on R&D. The team is part of Baidu’s newly created Autonomous Driving Unit (ADU), and Baidu plans to grow the team to over 100 researchers and engineers by end-2016. Baidu also plans to hire experts from the automotive industry. The Silicon Valley team will be known as ADU-US, and will focus on areas integral to self-driving car development, including planning, perception, control and systems. ADU-US will work alongside Baidu's existing Silicon Valley-based teams, including Baidu Research Institute.

Baidu plans to test its autonomous vehicles in 10 cities in China.

In May 2016, Baidu signed an agreement with Wuhu city in Anhui province to jointly build an exclusive pilot area for driverless vehicles in the city. The first driverless bus route is 2km long and runs from Chery Auto’s plant to its R&D center. The pilot area will be expanded over five years and eventually passengers will be allowed.

In June 2016, Baidu signed an agreement with Shanghai International Automobile City (an industrial park) to develop a 2 square kilometre Auto Innovation Park to test SAE Level-4 driverless vehicles.

In July 2016, Baidu signed a deal with Wuzhen Tourist Co. in Zhejiang province to develop driverless car services to ferry tourists from their hotel to various tourist attractions.

In Sep 2016, Baidu received an autonomous vehicle testing permit by the California Department of Motor Vehicles to test the vehicle on the state’s public roads and demonstrate compliance with stricter US regulations.

What if Baidu succeeds?

It may be too soon to consider the potential financial benefits of Baidu’s autonomous public shuttles, mass produced electric vehicles, and autonomous driving open platform, given these products will likely take at least three years from today to reach initial commercialisation, and business models are still unclear at the moment.

But the market potential is clearly huge, in our view. In Exhibit 10, we illustrate the potential revenue scale for Baidu.

Exhibit 10: Potential revenues for Baidu’s autonomous vehicle platform in different scenarios Bear case Base case Bull case

Scenario Autonomous vehicles never take off because of technology and regulatory barriers.

Autonomous vehicles become ubiquitous and Baidu takes significant market share in China.

Autonomous vehicles become ubiquitous and Baidu takes significant global market share.

Addressable market None 25m new car sales in China in 2016E 75m new car sales globally in 2016E

Penetration of autonomous cars by 2026 0% 50% 50%

Baidu market share 0% 30% of China market 20% of global market

Number of cars using Baidu's system 0 3.75m in China 7.5m globally

Revenue per installation 0 USD1000 USD1000

Potential revenue 0 USD3.75b USD7.5b

as % of Baidu 2016E revenue 0% 35% 69%

Note: We exclude commercial vehicles from this analysis. Sources: BNP Paribas estimates; Statista; China Association of Automobile Manufacturers

82



To be conservative we assume the only business model Baidu has in future is selling its autonomous driving platform to automobile OEMs, and we exclude commercial vehicles and public transport applications. We also assume each installation costs USD1000, which is about the price of a built-in GPS navigation system today, even though an autonomous driving platform is clearly more complex and should cost much more.

The bear case (which is highly improbable in our view) is that autonomous vehicles fail to ever take off due to technology and regulatory barriers. In this case Baidu would generate zero revenue from its investments.

We think a likely base case is that autonomous vehicles become ubiquitous by 2026 (10 years from now), and 50% of new cars sold in China will be autonomous. If Baidu takes a 30% market share in China, it would generate additional revenues equivalent to 35% of our current 2016E revenue estimate.

If Baidu also takes substantial global market share as in our bull case (which is not impossible given there are less than 10 leading global players providing autonomous driving platforms currently, such as Google, Tesla and some major automobile OEMs), then the additional revenue contribution could be as high as 69% of our 2016E estimate, if Baidu takes a 20% global market share.

Valuation and estimate changes

Autonomous driving is just one application of Baidu’s AI technology. There is clearly a lot of potential for AI to eventually impact all aspects of our lives. We think it is likely that Baidu would benefit from multiple AI commercialization opportunities over the next decade, especially in the China market which is relatively difficult to penetrate for foreign tech companies, and where Baidu is the leading player in AI.

We have conservatively factored in a 3% increase (cRMB3b) in our revenue estimates from 2018E onwards to capture some potential commercialisation benefits of AI. As discussed above it is difficult to specifically quantify the opportunities, but: 1) the addressable market can be very large; 2) Baidu is in a leading position; and 3) if successful, AI could be a huge swing factor – our illustration on autonomous driving alone could increase revenues by 35-69% (we have not put this into our forecasts).

We have cut our 2016E operating margin by 1.3ppt, resulting in a cut in our net profit estimate by 10% for the year. This is due to applying more conservative assumptions on Baidu’s SG&A expenses in 2H16. Even though Baidu is actively managing its O2O investments, we think competition in the space remains high, particularly in the food delivery segment, and could lead to higher expenses than we previously expected. We have left our 2017E estimates largely unchanged, but raised our 2018E operating profit and net profit estimates by 11% and 10% respectively. This is due to the additional revenues we factor in for potential AI commercialisation.

Exhibit 11: Old vs new forecasts

---------------------- 2016E --------------------- ---------------------- 2017E ---------------------- ---------------------- 2018E ----------------------

Old New Change Old New Change Old New Change


Revenue 72,791 72,385 (0.6) 93,380 93,685 0.3 111,402 114,737 3.0

Operating profit 12,171 11,173 (8.2) 17,714 17,847 0.8 24,708 27,523 11.4

Operating margin (%) 16.7 15.4 (1.3) ppt 19.0 19.0 0.1 ppt 22.2 24.0 1.8 ppt

Net profit 10,875 9,807 (9.8) 17,318 17,416 0.6 23,597 25,862 9.6

Net profit margin (%) 14.9 13.5 (1.4) ppt 18.5 18.6 0.0 ppt 21.2 22.5 1.4 ppt


83



Our target price is raised 18% from USD195 to USD230 as we adjust our 2017-18E EPS estimates by +1%/+10%. Our DCF assumptions are unchanged (3% terminal growth, 10.4% WACC). Our target price implies a 2017E P/E multiple of 31x.

While we value Baidu’s businesses as a whole using a DCF methodology, it is possible to work out the value of Baidu’s core search business as implied by the current market price. Exhibit 13 shows that if we strip out the market values of Baidu’s stakes in Ctrip and iQiyi, as well as its net cash balance, the core search business is trading at just 11.3x 2017 P/E. We think this is an attractive price for a business that grew revenues 27% y-y in 2015 and 48% y-y in 2014.

Exhibit 12: Changes to our revenue and margin forecasts

Sources: Baidu; BNP Paribas estimates

Exhibit 13: Market implied valuation of Baidu’s core search business

Unit Value Remarks

Baidu market capitalization USD b 62.1 As of 29 September 2016

Value of Ctrip stake USD b 5.2 22% stake in Ctrip, as of 29 September 2016

Value of iQiyi USD b 2.8 Buyout offer price by Robin Li and Yu Gong, which was withdrawn

Net cash USD b 5.5 As of 2Q16, converted using USD:RMB 6.66

Baidu core search business value USD b 48.6

2017 total revenue USD b 14.1 BNPPe, converted using USD:RMB 6.66

Core search revenue USD b 11.0 BNPPe, 78% of total revenues are from search and Baidu web properties

Core search margin % 48.8 As of 2Q16

2017 core search operating profit USD b 5.4 2017 core search net profit USD b 4.3 less 20% tax

Implied 2017 P/E for core business x 11.3

Note: Prices as of 29 September 2016 Sources: Bloomberg; BNP Paribas estimates

0

5

10

15

20

25

30

35

40

2015 2016E 2017E 2018E 2019E 2020E

(%) New revenue growth Old revenue growth

New EBIT margin Old EBIT margin

84



Exhibit 14: P/E band (12-month forward)

Note: 2015 earnings per ADS are adjusted for the after-tax gain from Ctrip-Qunar share swap transaction Sources: Bloomberg; BNP Paribas estimates

0

50

100

150

200

250

300

350

Aug-

05

Feb-

06

Aug-

06

Feb-

07

Aug-

07

Feb-

08

Aug-

08

Feb-

09

Aug-

09

Feb-

10

Aug-

10

Feb-

11

Aug-

11

Feb-

12

Aug-

12

Feb-

13

Aug-

13

Feb-

14

Aug-

14

Feb-

15

Aug-

15

Feb-

16

Aug-

16

(USD)

35x

50x

28x

16x

85



Financial statements Baidu



Revenue 49,052 66,382 72,385 93,685 114,737

Cost of sales ex depreciation (18,885) (27,458) (35,737) (46,775) (56,676)


Operating costs (13,391) (21,391) (19,108) (21,785) (21,668)

Operating EBITDA 16,776 17,532 17,540 25,125 36,393Depreciation (3,972) (5,861) (6,367) (7,278) (8,870)


Operating EBIT 12,804 11,672 11,173 17,847 27,523Net financing costs 1,671 26,235 1,906 4,220 5,666

Associates 0 0 0 0 0

Recurring non operating income 0 0 0 0 0


Profit before tax 14,475 37,907 13,078 22,067 33,189Tax (2,231) (5,474) (3,261) (4,413) (6,638)

Profit after tax 12,243 32,432 9,817 17,653 26,551Minority interests 944 1,232 (10) (238) (689)



Reported net profit 13,187 33,664 9,807 17,416 25,862Non recurring items & goodwill (net) 0 0 0 0 0

Recurring net profit 13,187 33,664 9,807 17,416 25,862

Per share (RMB)

Recurring EPS * 37.61 96.40 28.31 50.28 74.66

Reported EPS 37.61 96.40 28.31 50.28 74.66

DPS 0 0 0 0 0

Growth

Revenue (%) 53.6 35.3 9.0 29.4 22.5


Operating EBIT (%) 14.4 (8.8) (4.3) 59.7 54.2

Recurring EPS (%) 25.1 156.3 (70.6) 77.6 48.5

Reported EPS (%) 25.1 156.3 (70.6) 77.6 48.5





Net margin (%) 26.9 50.7 13.5 18.6 22.5




Inventory days 0.0 0.0 0.0 0.0 0.0

Debtor days 21.9 20.9 20.7 19.1 19.6

Creditor days 196.4 204.7 190.4 174.1 180.4

Operating ROIC (%) (1,245.4) (423.2) (290.0) (420.0) (493.9)

ROIC (%) 57.0 29.7 19.9 30.7 46.2

ROE (%) 29.3 51.1 11.4 17.2 20.7



Online marketing service 48,487 64,037 66,511 86,929 104,305

Others 565 2,345 5,875 6,756 10,431

86



Financial statements Baidu




Depreciation 3,972 5,861 6,367 7,278 8,870

Associates & minorities (944) (1,232) 10 238 689

Other non-cash items 0 0 0 0 0

Recurring cash flow 16,216 38,293 16,185 24,932 35,421Change in working capital 3,726 4,991 1,349 4,787 4,732


Capex - new investment (6,391) (7,722) (8,385) (10,852) (13,290)

Free cash flow to equity 13,550 35,562 9,149 18,867 26,862Net acquisitions & disposals 0 0 0 0 0

Dividends paid 0 0 0 0 0

Non recurring cash flows (2,533) (34,049) (2,000) (2,000) (2,000)

Net cash flow 11,017 1,513 7,149 16,867 24,862Equity finance 706 2,711 1,561 2,245 3,148

Debt finance 8,194 9,249 (4,757) (4,785) (6,374)

Movement in cash 19,918 13,473 3,954 14,327 21,636

Per share (RMB)


FCF to equity per share 38.65 101.83 26.41 54.47 77.55




Net working capital (9,929) (14,920) (16,269) (21,056) (25,788)Tangible fixed assets 8,705 10,627 12,856 15,971 19,728

Operating invested capital (1,223) (4,293) (3,413) (5,085) (6,060)Goodwill 17,419 15,396 15,546 15,696 15,846

Other intangible assets 3,574 3,335 3,123 3,582 4,245

Investments 2,879 37,959 39,959 41,959 43,959

Other assets 976 2,450 2,450 2,450 2,450

Invested capital 23,625 54,846 57,664 58,601 60,440Cash & equivalents (58,084) (68,025) (71,819) (85,758) (106,555)

Short term debt 25,767 35,017 30,260 25,475 19,102


Net debt (32,316) (33,009) (41,559) (60,283) (87,453)Deferred tax 0 0 0 0 0

Other liabilities 1,396 3,622 3,622 3,622 3,622

Total equity 51,526 80,256 91,614 111,037 139,357

Minority interests 2,980 3,960 3,970 4,208 4,897


Per share (RMB)



Financial strength

Net debt/equity (%) (59.3) (39.2) (43.5) (52.3) (60.6)

Net debt/total assets (%) (32.4) (22.4) (26.7) (33.8) (41.9)

Current ratio (x) 1.5 1.3 1.4 1.7 2.0

CF interest cover (x) n/a n/a n/a n/a n/a


Valuation 2014A 2015A 2016E 2017E 2018E

Recurring P/E (x) * 32.3 12.6 42.9 24.2 16.3


Reported P/E (x) 32.3 12.6 42.9 24.2 16.3

Dividend yield (%) 0.0 0.0 0.0 0.0 0.0

P/CF (x) 26.3 11.1 26.0 16.9 11.9

P/FCF (x) 31.4 11.9 46.0 22.3 15.7

Price/book (x) 8.3 5.3 4.6 3.8 3.0


EV/EBITDA (x) 23.7 22.6 21.9 14.5 9.3



87

Tung Thi h El ectr onic 3552 TT BNP PARIBAS Laura Chen

30 SEPTEMBER 2016 TAIWAN / CAPITAL GOODS

TUNG THIH ELECTRONIC 3552 TT

Continues to benefit from ADAS KEY STOCK DATA


Autonomous driving: TTE is set to capitalise on the rising trendConsumers are demanding safer cars, and new governmentmandates are increasingly specifying expanded safety features, such as Advance Driver Assistance Systems (ADAS). Strategy Analyticsforecasts global ADAS shipments will see a 20%+ CAGR over 2014-19. We believe TTE, a tier-I auto-parts maker and leading globalprovider of sensors/cameras, parking assist, autonomous parking-related products and TPMS, should benefit from this growing trend.

A weak 3Q16 before a 4Q pick-upWe expect 3Q results to miss initial guidance, for the secondconsecutive quarter. We attribute this to an inventory issue and SUVcapacity restructuring at one of its tier I customers (Changan), whichled to slower-than-expected shipments. However, we expectshipments to recover in 4Q16, thanks to the launch of a newmodification. We also expect demand for Great Wall’s SUV andD2XX to remain solid in 2016, and that more new model launcheswill drive up earnings from various customers in 2017-2018.

Potential long-term benefit from content upgrade in low-end PVThe overall average parking radar penetration rate reached 74.5% inJuly, and was in excess of 50% in each vehicle price range in China,while the overall average rear-view image penetration rate was42.9%. We still see upside potential for parking radar and rear-viewimage equipment, as intensified competition should prompt Chinaauto OEMs to upgrade standard equipment even in entry level PV,thus benefiting cost-competitive auto parts suppliers such as TTE.

Maintain BUY rating with new TP of TWD527We lower 2016E/17E/18E EPS by 19.8%/17.7%/14.8%, reflectingdisappointing sales and FX losses in 2Q16 and further earningsweakness in 3Q16 due to slow seasonality and weak demand; thisleads us to be more cautious in our forecasts for the later years. Ournew DCF-based TP equates to 23.8x 2017E P/E, lower than TTE’shistorical forward P/E peak of 31.6x (2009-current) and higher thanglobal peers’ average of 19.5x in 2017, (based on Bloombergconsensus). As TTE has a superior ROE to these global peers(36.5% vs 19.4% on average), we believe it warrants a highervaluation; we expect TTE to deliver at least 50% NI growth in 2016.

Laura Chen [email protected]

+852 2825 1280

Jimmy Huang [email protected] +852 2825 1124

BUY UNCHANGED

TARGET PRICE TWD527.00

CLOSE TWD448.00

UP/DOWNSIDE +17.6%

PRIOR TP TWD657.00

CHANGE IN TP -19.79%


TARGET PRICE (%) (1.6) POSITIVE 8

EPS 2016 (%) 5.0 NEUTRAL 3

EPS 2017 (%) (4.3) NEGATIVE 0

2015A 2016E 2017E 2018E

Revenue 7,039 10,212 13,819 18,715

Rec. net profit 790 1,380 1,860 2,525

Recurring EPS (TWD) 9.39 16.40 22.10 30.00

Prior rec. EPS (TWD) 9.39 20.45 26.85 35.20

Chg. In EPS est. (%) 0.0 (19.8) (17.7) (14.8)

EPS growth (%) 63.0 74.7 34.8 35.7

Recurring P/E (x) 47.7 27.3 20.3 14.9

Dividend yield (%) 1.2 2.1 2.8 3.8

EV/EBITDA (x) 30.1 17.2 13.5 10.0

Price/book (x) 10.2 8.3 6.7 5.3

Net debt/Equity (%) 11.0 (2.8) 0.2 1.1

ROE (%) 22.6 33.4 36.5 39.6


Absolute (%) (3.9) (5.2) 113.8

Relative to country (%) (5.5) (12.2) 87.6

Next Results



Free float (%) 72

Major shareholder Chen Hsin Chung (5%)

12m high/low (TWD) 585.00/225.00


ADR ticker -


Issued shares (m) 84

YE Dec (TWD m)

November 2016

0

80

160

240

320

200

300

400

500


Tung Thih Electronic Rel to Taiwan TAIEX Index(TWD) (%)

88

Tung Thih Electronic 3552 TT Laura Chen


Investment thesis We reiterate our BUY rating with a new DCF-based target price of TWD527 as we roll over our valuation to 2017E and incorporate our estimate changes. Our TP equates to 23.8x 2017E P/E, lower than the historical forward P/E peak of 31.6x but far above the global peer average of 19.5x. YTD, the share has outperformed TAIEX by 21.9%, yet we think it remains attractive at current levels.

We remain long-term positive on TTE out to 2018E: 1) despite a hiccup in 2Q16-3Q16 shipments due to internal re-adjustments following inventory issues and a new models transition at tier-I customers, we believe TTE will regain momentum; its tier-I auto OEM inventory issues have eased and new models will launch in 4Q16; 2) order visibility for the next three to five years suggests solid shipment growth of 70%-80% y-y, as auto models require more than two years for joint project development with auto OEMs; and 3) we expect improving economies of scale following initial R&D investment. Despite securing orders from existing China OEM partners and penetrating new customers, TTE may face increasing competition. However, we remain confident given TTE’s leading position on product quality, pricing and service.

Catalyst We expect near-term share price weakness in anticipation of weak 3Q16 results. Yet, with key clients starting to launch new models in 4Q16, timed to meet demand prompted by the termination of a tax subsidy by the end of 2016, we expect China auto demand to start regaining momentum in 4Q16.

Risk to our call Downside risks include: 1) a greater-than-expected slowdown in China’s auto market, especially in SUVs amid a slowing economy; 2) slower-than-expected adoption of D2XX models, and 3) FX volatility, especially related to RMB/USD.


Tung Thih Electronics (TTE), the third largest parking radar manufacturer after Denso and Valeo, is a tier-1 OE supplier, providing automobile electronics and exporting products to 18 countries. Products include parking aid systems (PAS), CMOS, remote key entry, and others.

(TWD m) 2014 2015 2016E 2017E 2018E

PAS 3,702 4,434 5,986 8,081 10,910

CMOS&MDS 617 1,478 2,476 3,342 4,512

RKE 771 915 1,318 1,779 2,402

TPMS 51 141 368 551 827

Others 70 64 64 64

Key executives Joined Title

Chen Hsin-Chung 1979 Chairman Chen Shun-Jung 2008 Chief Technology Officer Li Chih-Liang 2009 China Division President

http://www.tungthih.com/TW/index.asp Sources: Tung Thih Electronic; BNP Paribas estimates


-------- Bull -------- -------- Base -------- -------- Bear --------

Year-end 31 Dec 2016E 2017E 2016E 2017E 2016E 2017E

Revenues (TWD m) 10,722 14,510 10,212 13,819 9,701 13,128

Changes (%) 5% 5% - - -5% -5%

Margin (%) 33.5% 32.6% 31.9% 31.1% 30.3% 29.5%

Changes (%) 5% 5% -5% -5%

Net profit (TWD m) 1,714 2,300 1,380 1,860 1,063 1,441

Changes (%) 24% 24% -23% -23%

Source: Tung Thih Electronic Source: BNP Paribas estimates

Event calendar

Date Event

16 Sep 2016 The 66th IAA (Internationale Automobil-Ausstellung) Commercial Vehicles Show

16 Nov 2016 3Q16 earnings release

The key earnings driver is margin.

We estimate every 5ppt increase in gross margin increases net profit 24% in 2016E /2017E, all else being equal.

We estimate every 5ppt decrease in margin decreases net profit 23% in 2016E/ 2017E, all else being equal.

PAS - 63.0%

CMOS&MDS - 21.0%

RKE - 13.0%

TPMS - 2.0%

Others - 1.0%

89



Autonomous driving is a key goal for OEMs, international tech giants, and auto parts plyers

Autonomous drive vehicles quickly appear to be becoming a key goal for global OEMs and leading technology companies. In China, companies including Baidu, FAW, SAIC, BAIC, Chery, and Changan have shown an interest in developing their own models, targeting mass production in FY21 at the earliest. The benefits of autonomous driving include reducing accidents and improving vehicular access for senior and disabled drivers. Although China may be behind other countries in terms of government regulation and technology know-how in this area, we expect the Chinese government to announce ADAS-related policy by mid-FY17, and for this to boost local demand.

Autonomous drive vehicles utilise various equipment and IT systems to replace the functions of the human brain, eyes, ears and hands. ADAS comprises a number of systems, of which we think park assist (PA) is most common, followed by lane departure warnings (LDW), blind-spot detection (BSD) and forward collision warning (FCW). Connected car systems offer functions like in-vehicle services (navigation, smart-device access, entertainment, etc), vehicle management (remote access to locks, horns, etc) and, most importantly, vehicle-to-vehicle (V2V) communication.

We believe TTE is preparing well for the trend of autonomous driving, given it is spending more on R&D and management has introduced plans for future products, from 2020-2030, including integrated systems in auto driving and auto parking assistance and adaptive cruise control. Taking auto parking as an example, TTE products currently help drivers by taking control of the steering wheel. Management expects to help drivers by taking control of the brake and accelerator in 2018, eventually eliminating the need for a driver to stay in the car when parking in 2020. We believe as autonomous driving develops, entry barriers will become higher as the amount of R&D needed and track record of safety increases.

China auto market end-demand remains healthy

According to CAAM, China PV shipments grew 26.6% y-y in August 2016, driven by shipment growth in MPV (35.9% y-y) and SUV (44.3%). In 8M16, China shipped 14.4m PV, up 12.7% y-y, with SUV and MPV shipments up 43.9% and 21.6% y-y respectively.

We believe China’s auto market will remain healthy in 2016, as we expect the expiration of a tax subsidy by the end of 4Q16 to stimulate demand, in addition to 2017 new models launches by OEMs. We estimate 2016 PV shipments of 22.6m, up 7.1% y-y, with SUV/MPV shipments up 35% and 7% y-y, respectively.

Exhibit 1: China PV shipment y-y growth

Sources: CAAM; BNP Paribas

(10)

(5)

0

5

10

15

20

25

30

Jan-

14Fe

b-14

Mar

-14

Apr-1

4M

ay-1

4Ju

n-14

Jul-1

4Au

g-14

Sep-

14O

ct-1

4N

ov-1

4D

ec-1

4Ja

n-15

Feb-

15M

ar-1

5Ap

r-15

May

-15

Jun-

15Ju

l-15

Aug-

15Se

p-15

Oct

-15

Nov

-15

Dec

-15

Jan-

16Fe

b-16

Mar

-16

Apr-1

6M

ay-1

6Ju

n-16

Jul-1

6Au

g-16

(%)

90



Exhibit 2: China PV market breakdown


Exhibit 3: China PV OEM target achievement rate in 1H16 versus in 1H15

2016 shipment target 1H16 shipment Achievement rate 2015 shipment target 1H15 shipment Achievement rate

(‘000 units) (‘000 units) (%) (‘000 units) (‘000 units) (%)

Local Changan 830 564 67.9 750 625 83.3

Jianghuai 600 334 55.6 300 161 53.7

Geely 600 288 47.9 450 255 56.7

Greatwall 950 394 41.5 850 363 42.7

Chery 600 222 37.0 400 230 57.4

Japanese brand JV GAC Toyota 440 274 62.3 400 193 48.4

DF Honda 450 248 55.1 500 168 33.6

FAW Toyota 650 323 49.7 610 270 44.3

DF Nissan 1080 478 44.3 1000 441 44.1

GAC Honda 620 209 33.7 520 245 47.0

German brand JV FAW VW 1820 886 48.7 1850 804 43.5

SH VW 2000 969 48.4 1900 939 49.4

U.S. brand JV SH-GM 1870 828 44.3 1900 789 41.5

Changan Ford 1000 424 42.4 1000 430 43.0

Other JVs Beijing Hyundai 1120 523 46.7 1160 450 38.8

Dongfeng Yueda Kia 680 286 42.0 600 303 50.5

DF Peugeot-Citroen 765 286 37.4 800 355 44.3

Sources: CAAM; Sohu; Company data; BNP Paribas

69 70 74 75 75 72 69 70 69 67 63 55 51

3 4 4 4 3 2 3 3 3 7 10 10

11

5 5 5 6 7 6 10 11 13 17 21 29 35 23 21 18 16 16 19 18 16 15 9 7 5 4

0

10

20

30

40

50

60

70

80

90

100

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 8M16

(%) Sedan MPV SUV Crossover

91



Starting from April 2016, SUV growth has continued to support the overall China auto market, with overall SUV shipments growing 31.6% y-y in April, whereas overall auto shipments that month were up just 6.5% y-y.

In 8M16, China SUV shipments grew 44.0% y-y, to a total of 5.0m units. SUV demand remained solid even in the face of stronger competition from foreign JV brands against local names over the period. Some JV brands started to provide low-end SUV models, including FAW-Volkswagen, Hyundai, and KIA. We also saw ranking changes among the top-selling SUV models as JV brands increased market shares.

Before April 2016, five of the top 10 bestselling SUVs in China were local brands, including Great Wall, Changan, SAIC-GM-Wuling, GAC Motors, and JAC Motors. While market competition has intensified and some players have been pushed out of this list, Great Wall, GAC motors, and SAIC-GM-Wuling remain. Meanwhile, Japanese and Korean JV brands have started to catch up with local brands, and SH-GM, SH-VW, Beijing Hyundai and Gunagqi Honda appeared on the SUV top 10 bestselling models list in August.

SUV shipment growth exceeded all other segments in 1H16. Within that, JV brands gained momentum and saw their shipment growth exceed that of local brands. Although JV brands are increasing market share, we believe the stronger shipment y-y growth was due to a lower base.

Exhibit 4: China FAI-Automotive manufacturing and y-y Exhibit 5: China PV OEMs UT rate

Source: Bloomberg Sources: CAAM; BNP Paribas

Exhibit 6: Vehicle inventory alert index

Source: CADA

(10)

(5)

0

5

10

15

20

25

30

35

0

200

400

600

800

1,000

1,200

1,400

2010 2011 2012 2013 2014 2015 5M16

(y-y %)(RMB b) China FAI-Automotive manufacturing (LHS)Change (RHS)

60

65

70

75

80

85

90

95

100

105

2012 2013 2014 2015

(%)

40

45

50

55

60

65

70

Mar

-15

Apr-1

5

May

-15

Jun-

15

Jul-1

5

Aug-

15

Sep-

15

Oct

-15

Nov

-15

Dec

-15

Jan-

16

Feb-

16

Mar

-16

Apr-1

6

May

-16

Jun-

16

Jul-1

6

Aug-

16

(%)

92



China ADAS penetration remains low in low-priced PVs

ADAS cost has not yet reached the sweet spot, and it still may take years before it reaches that point. The majority of PV sold in China are still mid- to low end vehicles; price points below RMB200k account for 67% of the entire China PV market (as of July 2016), leading us to believe that ADAS applications will gradually move down from high-end models to mid-end in order to attract more customers with content upgrades. For vehicles priced below RMB100k, penetration of parking radar, rear-view image equipment and TPMS remains the lowest among all price brackets, at 56.3%, 24.4%, and 8.4%, respectively. The overall average parking radar penetration rate reached 74.5% in July, and was in excess of 50% in each vehicle price range in China, while the overall average rear-view image penetration rate was 42.9%.

We thus assume the rear-view image penetration rate of PV priced below RMB200k is lower than 50%, while we estimate that rear-view image penetration on PV priced below RMB500k also has not yet reached 60%. Consequently, we believe rear-view image equipment shipments still have room to grow, and expect cost-competitive manufacturers such as TTE to still benefit from this auto parts content upgrade trend.

The average TPMS penetration rate is 34.7%. We believe there still is scope for demand upside. We expect China’s authorities to make TPMS a mandatory spec by 2017, as has already been proposed by the National Technical Committee. Thus, we

Exhibit 7: China SUV shipment y-y growth


Exhibit 8: China SUV OEMs with shipment growth above average in 1H15

Exhibit 9: China SUV OEMs with shipment growth above average in 1H16

Source: CAAM; BNP Paribas Research Source: CAAM; BNP Paribas Research

0

10

20

30

40

50

60

70

80

Jan-

14Fe

b-14

Mar

-14

Apr-1

4M

ay-1

4Ju

n-14

Jul-1

4Au

g-14

Sep-

14O

ct-1

4N

ov-1

4D

ec-1

4Ja

n-15

Feb-

15M

ar-1

5Ap

r-15

May

-15

Jun-

15Ju

l-15

Aug-

15Se

p-15

Oct

-15

Nov

-15

Dec

-15

Jan-

16Fe

b-16

Mar

-16

Apr-1

6M

ay-1

6Ju

n-16

Jul-1

6Au

g-16

(%)

0 150 300 450 600

Total

Beijing Benz

Great Wall Motor

Sichuan Auto

Beijing Auto Works

South East Motor

Brilliance Jinbei

Hawtai Auto

Dongfeng PSA

SAIC GM

Jiangnan Auto

Dongfeng Yulon

Haima Auto

Changan Auto

Zhengzhou Nissan

Jianghuai Auto

BAIC Yinxiang

(%) 0 50 100 150 200 250 300

TotalChangan Suzuki

GAC HondaDongfeng Honda

BYDBeijing HyundaiBrilliance Jinbei

Dongfeng Yueda KiaHaima AutoSAIC Motor

Beijing BenzGAC Toyota

Jiangling AutoBAIC Guangzhou

Jiangnan AutoSichuan Auto

Qoros AutomotiveGAC MotorBeiqi Foton

(%)

93



believe TTE should benefit from its solid China customer base, although current sales contributions from TPMS remain limited.

Diversify customer mix with more models and products

TTE currently supplies the Jianging Landwind S7, with TWD15,000 of content per vehicle. However, we believe that if the new model Landwind S5 is supplied by TTE, content per vehicle is likely to be less than TWD10,000, as this is a lower-spec, lower priced vehicle. In addition, two or three new Jiangling models will also be supplied by TTE in 2017, equipped with AVM (Around View Monitor) and parking radar.

For Changan, TTE says it will start supplying AVM between 4Q16 and 1Q17, and in 2H17 for GWM. For SH-GM, TTE will add three new models. It will start supplying parking camera and auto parking assistance for one of these – the GL8 – by the end of 2016, and in 2017 for the other two models.

Over the first eight months of 2016, TTE management reports growing revenue contributions from India Suzuki, Dongfeng Motor (excluding JV) and Beijing Hyundai. TTE management said it is also expecting to supply Beijing Hyundai MISTRA in 4Q16 or 1Q17.

Although TTE will not be supplying the second generation D2XX platform models after 2020 or 2021, management is not worried about idle capacity; the manufacturing line can be easily adjusted for other products as the company diversifies its customers mix, with high-end products such as LDW (Lane Departure Warning) expected to make more meaningful contributions from 2018, given that the cost to produce in-house is still high for OEMs at the current level.

Exhibit 10: Parking radar, rear-view image, and TPMS penetration rates in China PVs among different pricing ranges

Sources: Sohu; BNP Paribas

Exhibit 11: 2015 revenue breakdown by client Exhibit 12: 8M16 revenue breakdown by client

Source: TTE Source: TTE

0

10

20

30

40

50

60

70

80

90

100

BelowRMB 100K

RMB100K-200K

RMB200K-300K

RMB300K-500K

Above500K

Average

(%) Parking Radar Rearview image TPMS

SH-VW4.0% DFN

6.0%

SH-GM9.0%

GWM13.0%

Changan11.0%

Jiangling8.0%

SAIC-GM-Wuling6.0%

Others43.0%

SH-GM10.0%

GWM10.0%

Changan11.0%

Jiangling13.0%

SAIC-GM-Wuling7.0%

Others49.0%

94



2016-2018E earnings revisions

We lower our 2016 revenue forecast by 17.0% to reflect slower sales in 2Q16 and 3Q16E. We also lowered our revenue forecasts for 2017-2018 by 10.6% and 5.4%, as we are now are more conservative due to 2016E’s weakness. However, we still believe TTE could deliver 45.1%/35.3%/35.4% sales growth in 2016/2017/2018 as tier-I customers start to launch new models in 4Q16, orders increase from new and existing customers, and as content ASP per vehicle increases. In addition, order visibility for the next three to five years suggests solid shipment growth of 70-80% y-y, as auto models require more than two years for joint project development with auto OEMs.

We also expect both gross margin and operating margin to continue to expand to 31.9% and19.4%, respectively in 2016 from 2015’s 30.6% and 15.5%, respectively thanks to better product mix and cost control. We expect to see a 2016-18 earnings CAGR of 35.2%.

Exhibit 13: Revenue breakdown by product

[PAS :Parking Aid System; RKE: Remote Key Entry; TPMS: Tire Pressure Monitoring System Source: TTE

Exhibit 14: 2016-2018E earnings revisions

------------------ 2016E ------------------ ------------------ 2017E ------------------ ------------------ 2018E ------------------

New Old Diff. New Old Diff. New Old Diff.

(TWD m) (TWD m) (%) (TWD m) (TWD m) (%) (TWD m) (TWD m) (%)

Total sales 10,212 12,306 (17.0) 13,819 15,463 (10.6) 18,715 19,789 (5.4)

Cost of Goods Sold (6,958) (8,512) (9,524) (10,694) (12,880) (13,677) Gross profit 3,254 3,794 (14.2) 4,295 4,770 (10.0) 5,835 6,112 (4.5)

Operating expenses (1,274) (1,516) (1,741) (1,757) (2,356) (2,166) Operating profit 1,980 2,278 (13.1) 2,554 3,013 (15.2) 3,479 3,947 (11.8)

Total non-operating profit (112) 40 (30) 14 (54) 22 Pre-tax profit 1,868 2,317 (19.4) 2,523 3,027 (16.6) 3,425 3,968 (13.7)

Income tax (488) (596) (663) (767) (900) (1,006) Net profit 1,380 1,721 (19.8) 1,860 2,260 (17.7) 2,525 2,962 (14.8)

Outstanding shares (m) 84 84 84 84 84 84 EPS (TWD) 16.40 20.45 (19.8) 22.10 26.85 (17.7) 30.00 35.20 (14.8)

Margins (%)

Gross margin 31.9 30.8 1.0 31.1 30.8 0.2 31.2 30.9 0.3

Operating margin 19.4 18.5 0.9 18.5 19.5 (1.0) 18.6 19.9 (1.4)

Net margin 13.5 14.0 (0.5) 13.5 14.6 (1.2) 13.5 15.0 (1.5)


0

10

20

30

40

50

60

70

80

90

100

1Q15 2Q15 3Q15 4Q15 1Q16 2Q16

(%) PAS CMOS&MDS RKE TPMS Others

95



Valuation

We maintain our BUY rating with a new DCF-based TP of TWD527, which implies 23.8x 2017E P/E, lower than TTE’s historical forward P/E peak of 31.6x (2009-current). We believe a 2017E ROE of 36.5% versus the global peer average of 19.4% and 2016-18E earnings CAGR of 35.2% will let TTE trade at a premium to its global peers.

Exhibit 15: Trailing P/E versus forward P/E Exhibit 16: Forward P/E versus earnings growth

Sources: TEJ; BNP Paribas estimates Sources: TEJ; BNP Paribas estimates

Exhibit 17: Peer valuations Company BBG code Price Mkt. cap. --------- P/E --------- ------- P/B ------- ------- ROE ------- ---- Div. yield ---- --- EPS growth ---

2016E 2017E 2016E 2017E 2016E 2017E 2016E 2017E 2016E 2017E

(LC) (USD m) (x) (x) (x) (x) (%) (%) (%) (%) (%) (%)

Tung Thih* 3552 TT 448 1,197 27.3 20.3 8.3 6.7 33.4 36.5 2.1 2.8 74.7 34.8

Taiwan Auto Parts Player

Cub elecparts 2231 TT 343.0 914.2 21.4 17.3 8.8 7.3 45.6 46.5 2.9 3.5 19.1 23.5

Hota 1536 TT 152.5 1,141.8 27.6 21.8 8.0 6.8 31.4 34.4 2.3 2.9 19.6 26.4

Tong Yang 1319 TT 76.0 1,432.4 18.6 16.0 2.3 2.1 12.4 13.4 2.4 2.9 42.2 16.4

Iron Force 2228 TT 238.0 574.4 24.2 20.0 4.2 - 21.5 22.8 2.1 2.5 15.0 21.0

Hu Lane 6279 TT 155.5 481.4 17.9 15.2 3.4 3.2 20.4 22.1 3.8 4.4 5.2 17.7

BizLink 3665 TT 164.5 513.3 18.9 16.2 3.2 2.9 18.7 19.5 3.6 4.3 10.0 16.6

Hiroca 1338 TT 123.0 328.6 12.9 11.2 1.5 1.4 12.4 13.5 3.9 4.6 5.7 15.0

Ta Yih 1521 TT 82.3 199.9 11.7 - 3.3 - 28.6 - - - 19.2 -

E-Lead 2497 TT 30.0 113.6 13.0 - - - - - - - 29.4 -

Summeko 2066 TT 120.5 140.2 16.3 15.0 3.8 - 24.5 - 4.8 - 4.1 8.9

Turvo 2233 TT 81.5 127.8 13.2 12.2 1.6 - 13.3 13.3 3.9 4.2 21.5 8.2

Lu Hai 2115 TT 62.1 147.4 14.7 13.1 2.1 2.0 14.5 15.8 4.5 5.0 25.8 12.3

Average 17.5 15.8 3.8 3.7 22.1 22.4 3.4 3.8 18.1 16.6

Global Auto Parts player

Denso 6902 JP 4,015.0 31,556.8 14.3 12.8 1.0 1.0 6.9 7.6 3.0 3.1 (8.9) 12.1

Valeo FR FP 51.1 13,659.5 13.8 12.1 3.0 2.6 22.5 22.2 2.3 2.5 19.3 13.8

Autoliv ALV US 104.2 9,188.0 15.6 14.3 2.5 2.3 15.9 16.1 2.3 2.4 28.6 9.5

Mobileye MBLY US 41.7 9,135.2 58.5 39.0 13.6 9.9 28.1 31.6 - - 129.7 50.1

Average 25.6 19.5 5.0 3.9 18.4 19.4 1.9 2.0 42.2 21.4

Market Consensus

Tung Thih 3552 TT 448.0 1,204.4 26.0 20.2 8.2 6.5 35.2 38.0 1.9 2.7 83.7 28.6

Average 19.9 17.1 4.4 4.0 22.0 22.6 2.9 3.2 27.6 18.7

Prices are as of 29 September 2016 Sources: Bloomberg; *BNP Paribas estimates

0

10

20

30

40

50

60

Jan-

09Ju

n-09

Nov

-09

Apr-1

0Se

p-10

Feb-

11Ju

l-11

Dec

-11

May

-12

Oct

-12

Mar

-13

Aug-

13Ja

n-14

Jun-

14N

ov-1

4Ap

r-15

Sep-

15Fe

b-16

Jul-1

6

(x) Trailing P/E Forward P/E

Median trailing P/E: 19.4x

Median forward P/E: 14.4x

(40)

(20)

0

20

40

60

80

100

0

5

10

15

20

25

30

35

Jan-

09

Jan-

10

Jan-

11

Jan-

12

Jan-

13

Jan-

14

Jan-

15

Jan-

16

(%)(x) Forward P/E (LHS)Earnings growth (RHS)

96



Exhibit 18: DCF valuation

2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031

No. of years 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Total sales (TWD m) 10,212 13,819 18,715 25,067 33,197 43,467 56,262 71,980 91,008 113,702 140,349 171,137 206,110 245,139 287,882 333,759

Growth (y-y %) 45 35 35 33.9 32.4 30.9 29.4 27.9 26.4 24.9 23.4 21.9 20.4 18.9 17.4 15.9

Operating profits (TWD m) 1,980 2,554 3,479 4,471 5,672 7,099 8,765 10,671 12,807 15,145 17,637 20,217 22,797 25,268 27,506 29,376

Operating margin (%) 19.4 18.5 18.6 17.8 17.1 16.3 15.6 14.8 14.1 13.3 12.6 11.8 11.1 10.3 9.6 8.8

Depreciation 215 260 316 313 310 307 304 301 298 295 292 289 286 283 280 278

EBITDA 2,195 2,814 3,796 4,784 5,982 7,406 9,069 10,972 13,105 15,439 17,929 20,506 23,083 25,551 27,786 29,654

EBITDA margin (%) 21.5 20.4 20.3 19.1 18.0 17.0 16.1 15.2 14.4 13.6 12.8 12.0 11.2 10.4 9.7 8.9

Working capital 3,162 4,341 5,948 7,967 10,551 13,815 17,882 22,877 28,925 36,138 44,607 54,392 65,508 77,912 91,497 106,078

% of sales 31 31 32 32 32 32 32 32 32 32 32 32 32 32 32 32

Net working capital (537) (1,179) (1,607) (2,019) (2,584) (3,264) (4,067) (4,995) (6,048) (7,213) (8,469) (9,785) (11,116) (12,405) (13,585) (14,581)

Capex (132) (500) (500) (525) (551) (579) (608) (638) (670) (704) (739) (776) (814) (855) (898) (943)

Tax paid (488) (663) (900) (1,341) (1,701) (2,130) (2,629) (3,201) (3,842) (4,543) (5,291) (6,065) (6,839) (7,580) (8,252) (8,813)

Other adjustment - - - - - - - - - - - - - - - -

Free Cash Flow 1,038 472 788 899 1,145 1,433 1,765 2,137 2,545 2,980 3,430 3,880 4,314 4,711 5,052 5,317

DCF Calculation 2017 NPV of cash flows 17,621 Perpetual growth (%) 3 PV of terminal value 25,469 Enterprise value 43,090 Terminal value as % of EV (%) 59 Net debt (1,343) Equity value 44,433 No. of shares (m) 84 DCF value per share 527.0 Fair value based on 32x peak P/E 707.3 Difference (%) (25) Current price (TWD) 448.0 Difference (%) 17.6 Implied P/E 23.8 WACC Calculation Equity Risk Free Rate (Rf) (%) 1 Equity risk premium (Erp) (%) 10 Beta 0.88

Equity required rate of return (Re) (%) 10.8 CAPM =Rf+B(Rm-Rf) (%) 9.7 % of equity 91.2

Debt Cost of debt (%) 0 Marginal tax rate (t) (%) 29 % of debt 8.8

WACC (%) 8.9


97



Financial statements Tung Thih Electronic

Sources: Tung Thih Electronic; BNP Paribas estimates

Profit and Loss (TWD m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E

Revenue 5,142 7,039 10,212 13,819 18,715

Cost of sales ex depreciation (3,421) (4,705) (6,743) (9,264) (12,564)


Operating costs (920) (1,062) (1,274) (1,741) (2,356)

Operating EBITDA 801 1,272 2,195 2,814 3,796Depreciation (143) (178) (215) (260) (316)


Operating EBIT 658 1,094 1,980 2,554 3,479Net financing costs 31 10 2 22 16

Associates 0 0 0 0 0

Recurring non operating income 11 (55) (114) (52) (70)


Profit before tax 700 1,049 1,868 2,523 3,425Tax (224) (258) (488) (663) (900)

Profit after tax 476 790 1,380 1,860 2,525Minority interests 0 0 0 0 0



Reported net profit 476 790 1,380 1,860 2,525Non recurring items & goodwill (net) 0 0 0 0 0

Recurring net profit 476 790 1,380 1,860 2,525

Per share (TWD)

Recurring EPS * 5.76 9.39 16.40 22.10 30.00

Reported EPS 5.76 9.39 16.40 22.10 30.00

DPS 3.40 5.25 9.21 12.41 16.84

Growth

Revenue (%) 14.0 36.9 45.1 35.3 35.4


Operating EBIT (%) 41.5 66.3 81.1 29.0 36.3

Recurring EPS (%) 6.3 63.0 74.7 34.8 35.7

Reported EPS (%) 6.3 63.0 74.7 34.8 35.7





Net margin (%) 9.3 11.2 13.5 13.5 13.5




Inventory days 124.1 122.9 104.1 93.7 93.8

Debtor days 108.0 123.1 122.2 119.0 118.7

Creditor days 116.7 125.8 119.7 114.4 114.5

Operating ROIC (%) 15.3 22.6 36.9 40.1 42.2

ROIC (%) 14.2 19.7 31.8 36.2 38.8

ROE (%) 15.8 22.6 33.4 36.5 39.6


Revenue By Division (TWD m) 2014A 2015A 2016E 2017E 2018E

PAS 3,702 4,434 5,986 8,081 10,910

CMOS&MDS 617 1,478 2,476 3,342 4,512

RKE 771 915 1,318 1,779 2,402

TPMS 51 141 368 551 827

Others 0 70 64 64 64

98



Financial statements Tung Thih Electronic

Sources: Tung Thih Electronic; BNP Paribas estimates

Cash Flow (TWD m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E

Recurring net profit 476 790 1,380 1,860 2,525

Depreciation 143 178 215 260 316

Associates & minorities 6 16 (23) 0 0

Other non-cash items 7 5 (12) 0 0

Recurring cash flow 632 990 1,560 2,120 2,841Change in working capital (851) (266) (172) (986) (1,359)


Capex - new investment (337) (325) (272) (500) (500)

Free cash flow to equity (556) 399 1,116 634 982Net acquisitions & disposals 0 0 0 0 0

Dividends paid (249) (286) (444) (775) (1,044)

Non recurring cash flows (50) (54) (25) 0 0

Net cash flow (855) 59 647 (141) (62)Equity finance 316 (53) (109) 0 0

Debt finance 99 438 44 0 0

Movement in cash (440) 444 581 (141) (62)

Per share (TWD)


FCF to equity per share (6.73) 4.74 13.26 7.54 11.67

Balance Sheet (TWD m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E



Net working capital 2,219 2,485 2,657 3,643 5,002Tangible fixed assets 1,220 1,367 1,423 1,663 1,847

Operating invested capital 3,439 3,851 4,080 5,306 6,849Goodwill 0 0 0 0 0

Other intangible assets 75 74 81 81 81

Investments 21 0 35 35 35

Other assets 215 271 289 289 289

Invested capital 3,750 4,196 4,486 5,712 7,254Cash & equivalents (930) (1,375) (1,956) (1,815) (1,754)

Short term debt 1,345 1,784 1,828 1,828 1,828


Net debt 422 410 (127) 13 75Deferred tax 0 0 0 0 0

Other liabilities 55 61 60 60 60

Total equity 3,274 3,725 4,553 5,638 7,119

Minority interests 0 0 0 0 0


Per share (TWD)



Financial strength

Net debt/equity (%) 12.9 11.0 (2.8) 0.2 1.1

Net debt/total assets (%) 6.4 4.9 (1.3) 0.1 0.5

Current ratio (x) 1.6 1.5 1.5 1.6 1.6

CF interest cover (x) 8.0 n/a n/a n/a n/a


Valuation 2014A 2015A 2016E 2017E 2018E

Recurring P/E (x) * 77.8 47.7 27.3 20.3 14.9


Reported P/E (x) 77.8 47.7 27.3 20.3 14.9

Dividend yield (%) 0.8 1.2 2.1 2.8 3.8

P/CF (x) 58.6 38.1 24.2 17.8 13.3

P/FCF (x) (66.5) 94.5 33.8 59.4 38.4

Price/book (x) 11.5 10.2 8.3 6.7 5.3


EV/EBITDA (x) 47.4 30.1 17.2 13.5 10.0



99



Financial statementsWistron NeWeb

Profit and Loss (TWD m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E

Revenue 40,327 52,183 55,704 66,674 75,529Cost of sales ex depreciation (33,929) (44,153) (47,271) (56,434) (63,950)

Gross profit ex depreciation 6,398 8,030 8,433 10,241 11,579Other operating income 0 0 0 0 0Operating costs (3,712) (4,583) (4,729) (5,811) (6,705)

Operating EBITDA 2,686 3,447 3,704 4,429 4,874Depreciation (984) (1,120) (1,164) (1,174) (1,180)Goodwill amortisation 0 0 0 0 0

Operating EBIT 1,703 2,327 2,539 3,255 3,694Net financing costs (9) 15 (5) (4) (3)Associates 11 18 3 0 0Recurring non operating income 86 208 11 72 32Non recurring items 0 0 0 0 0

Profit before tax 1,780 2,550 2,545 3,323 3,723Tax (404) (577) (528) (665) (745)

Profit after tax 1,376 1,973 2,018 2,659 2,978Minority interests 0 0 0 0 0Preferred dividends 0 0 0 0 0Other items 0 0 0 0 0

Reported net profit 1,376 1,973 2,018 2,659 2,978Non recurring items & goodwill (net) 0 0 0 0 0


Per share (TWD)

Recurring EPS * 4.32 6.00 6.07 7.99 8.95Reported EPS 4.32 6.00 6.07 7.99 8.95DPS 2.94 2.70 3.82 3.91 5.15

Growth

Revenue (%) 10.0 29.4 6.7 19.7 13.3Operating EBITDA (%) 0.3 28.3 7.4 19.6 10.0Operating EBIT (%) (4.5) 36.7 9.1 28.2 13.5Recurring EPS (%) (11.9) 38.8 1.1 31.8 12.0Reported EPS (%) (11.9) 38.8 1.1 31.8 12.0


Gross margin inc depreciation (%) 13.4 13.2 13.0 13.6 13.8Operating EBITDA margin (%) 6.7 6.6 6.6 6.6 6.5Operating EBIT margin (%) 4.2 4.5 4.6 4.9 4.9Net margin (%) 3.4 3.8 3.6 4.0 3.9Effective tax rate (%) 22.8 22.8 20.7 20.0 20.0Dividend payout on recurring profit (%) 67.9 45.0 63.1 48.9 57.6Interest cover (x) 201.3 n/a 482.3 849.2 1,239.5Inventory days 40.9 41.0 41.6 34.6 31.8Debtor days 59.0 52.6 54.6 55.1 57.0Creditor days 68.5 64.0 60.7 51.0 50.2Operating ROIC (%) 25.2 29.4 27.8 0.0 (32.4)ROIC (%) 23.7 29.0 25.5 0.0 (30.4)ROE (%) 12.7 17.2 16.5 20.2 21.0ROA (%) 6.0 8.1 8.0 9.9 10.1*Pre exceptional pre-goodwill and fully diluted

Revenue By Division (TWD m) 2014A 2015A 2016E 2017E 2018E

Smart Home 26,585 34,705 37,308 42,235 47,900Auto-related 10,173 14,480 16,412 22,536 25,996Mobile-related 2,229 2,507 1,552 1,468 1,181Others 1,340 491 431 435 452

Source: Wistron NeWeb, BNP Paribas estimates

100



Financial statementsWistron NeWeb

Cash Flow (TWD m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E

Recurring net profit 1,376 1,973 2,018 2,659 2,978Depreciation 984 1,120 1,164 1,174 1,180Associates & minorities (11) (18) (3) 0 0Other non-cash items 0 0 (3) 0 0

Recurring cash flow 2,350 3,075 3,176 3,833 4,158Change in working capital (953) (804) (1,608) (1,321) (62)Capex - maintenance (445) (337) (345) (344) (407)Capex - new investment (1,039) (787) (805) (803) (950)

Free cash flow to equity (87) 1,147 418 1,364 2,739Net acquisitions & disposals 185 17 (28) (34) (38)Dividends paid (965) (888) (1,272) (1,301) (1,715)Non recurring cash flows (209) (121) 40 (131) 46

Net cash flow (1,077) 155 (842) (102) 1,033Equity finance 302 (644) 339 (612) 94Debt finance (1,644) 161 170 102 187

Movement in cash (2,419) (328) (333) (612) 1,314

Per share (TWD)

Recurring cash flow per share 7.38 9.35 9.55 11.52 12.50FCF to equity per share (0.27) 3.49 1.26 4.10 8.23

Balance Sheet (TWD m) Year Ending Dec 2014A 2015A 2016E 2017E 2018E

Working capital assets 12,372 13,356 15,027 16,995 19,124Working capital liabilities (10,090) (10,270) (10,333) (10,981) (13,048)

Net working capital 2,281 3,086 4,693 6,014 6,076Tangible fixed assets 5,233 5,253 5,257 5,247 5,441

Operating invested capital 7,514 8,339 9,950 11,261 11,517Goodwill 0 0 0 0 0Other intangible assets 0 0 0 0 0Investments 310 311 342 376 413Other assets 465 569 512 626 563

Invested capital 8,289 9,219 10,803 12,156 12,362Cash & equivalents (5,450) (5,122) (4,792) (4,287) (5,625)Short term debt 2,025 2,197 2,307 2,422 2,543Long term debt * 0 0 0 0 0

Net debt (3,425) (2,925) (2,485) (1,864) (3,082)Deferred tax - - - - -Other liabilities 479 467 527 514 579Total equity 11,235 11,677 12,761 13,507 14,864Minority interests 0 0 0 0 0


Per share (TWD)

Book value per share 34.18 35.52 38.36 40.60 44.69Tangible book value per share 34.18 35.52 38.36 40.60 44.69

Financial strength

Net debt/equity (%) (30.5) (25.0) (19.5) (13.8) (20.7)Net debt/total assets (%) (14.4) (11.9) (9.6) (6.8) (9.9)Current ratio (x) 1.5 1.5 1.6 1.6 1.6

Valuation 2014A 2015A 2016E 2017E 2018E

Recurring P/E (x) * 22.7 16.3 16.2 12.3 10.9Recurring P/E @ target price (x) * 25.9 18.7 18.5 14.0 12.5Reported P/E (x) 22.7 16.3 16.2 12.3 10.9Dividend yield (%) 3.0 2.8 3.9 4.0 5.3P/CF (x) 13.3 10.5 10.3 8.5 7.8P/FCF (x) (358.1) 28.1 78.0 23.9 11.9Price/book (x) 2.9 2.8 2.6 2.4 2.2Price/tangible book (x) 2.9 2.8 2.6 2.4 2.2EV/EBITDA (x) 10.7 8.5 8.1 6.9 6.1EV/EBITDA @ target price (x) 12.4 9.8 9.4 8.0 7.0


Source: Wistron NeWeb, BNP Paribas estimates


101



Financial statementsJoyson Electronics


Revenue 7,077 8,083 19,059 27,133 30,822Cost of sales ex depreciation (5,390) (5,995) (15,078) (21,624) (24,602)

Gross profit ex depreciation 1,687 2,088 3,980 5,509 6,219Other operating income - - - - -Operating costs (877) (1,162) (2,423) (3,507) (3,874)

Operating EBITDA 810 925 1,557 2,002 2,346Depreciation (303) (338) (340) (350) (350)Goodwill amortisation 0 0 0 0 0

Operating EBIT 508 587 1,217 1,652 1,996Net financing costs (63) (115) (300) (150) (150)Associates 1 73 29 20 20Recurring non operating income 14 87 77 68 40Non recurring items 0 0 0 0 0

Profit before tax 459 559 993 1,570 1,886Tax (109) (142) (254) (399) (479)

Profit after tax 350 417 739 1,171 1,407Minority interests (3) (17) (49) (44) (44)Preferred dividends 0 0 0 0 0Other items 0 0 0 0 0

Reported net profit 347 400 691 1,127 1,363Non recurring items & goodwill (net) 0 0 0 0 0

Recurring net profit 347 400 691 1,127 1,363

Per share (RMB)

Recurring EPS * 0.55 0.61 0.71 1.15 1.39Reported EPS 0.55 0.61 0.71 1.15 1.39DPS 0.24 0.35 0.14 0.23 0.28

Growth

Revenue (%) 15.9 14.2 135.8 42.4 13.6Operating EBITDA (%) 8.9 14.2 68.3 28.6 17.2Operating EBIT (%) 7.4 15.7 107.2 35.7 20.8Recurring EPS (%) 20.0 11.8 15.9 63.2 20.9Reported EPS (%) 20.0 11.8 15.9 63.2 20.9


Gross margin inc depreciation (%) 19.6 21.6 19.1 19.0 19.0Operating EBITDA margin (%) 11.4 11.4 8.2 7.4 7.6Operating EBIT margin (%) 7.2 7.3 6.4 6.1 6.5Net margin (%) 4.9 4.9 3.6 4.2 4.4Effective tax rate (%) 0.0 0.0 0.0 0.0 0.0Dividend payout on recurring profit (%) 44.3 57.3 20.0 20.0 20.0Interest cover (x) 8.3 5.9 4.3 11.5 13.6Inventory days 58.4 62.3 44.8 51.9 57.2Debtor days 51.5 52.5 40.7 50.4 56.4Creditor days 71.4 68.7 44.2 51.9 57.2Operating ROIC (%) 18.3 21.0 24.9 20.4 20.8ROIC (%) 13.8 16.6 11.2 8.6 9.5ROE (%) 14.7 12.9 8.3 8.4 9.5ROA (%) 6.9 6.0 5.5 4.7 5.2*Pre exceptional pre-goodwill and fully diluted


Electronic component 4,423 4,470 8,464 10,490 11,886Functional component 1,795 2,489 2,496 2,709 3,037Safety component (KSS) 0 0 6,450 11,804 13,130Industrial automation 362 557 850 1,160 1,508New energy 186 298 507 706 939Others 312 268 291 265 322

Source: Joyson Electronics, BNP Paribas estimates

102



Financial statementsJoyson Electronics


Recurring net profit 347 400 691 1,127 1,363Depreciation 303 338 340 350 350Associates & minorities 0 0 0 0 0Other non-cash items 160 277 0 0 0

Recurring cash flow 809 1,014 1,031 1,477 1,713Change in working capital (72) (429) (1,756) (1,432) (606)Capex - maintenance 0 0 0 0 0Capex - new investment (758) (853) (800) (900) (800)

Free cash flow to equity (21) (268) (1,526) (855) 306Net acquisitions & disposals 0 0 0 0 0Dividends paid (59) (154) (229) (138) (225)Non recurring cash flows (86) (990) (7,973) 0 0

Net cash flow (165) (1,411) (9,728) (993) 81Equity finance 1 1,147 8,600 0 0Debt finance 346 3,153 200 200 200

Movement in cash 182 2,889 (928) (793) 281

Per share (RMB)

Recurring cash flow per share 1.27 1.55 1.05 1.51 1.75FCF to equity per share (0.03) (0.41) (1.56) (0.87) 0.31


Working capital assets 2,230 3,561 6,731 9,241 10,337Working capital liabilities (1,752) (3,801) (5,215) (6,293) (6,783)

Net working capital 479 (241) 1,516 2,948 3,554Tangible fixed assets 2,439 2,921 5,581 6,131 6,581

Operating invested capital 2,918 2,681 7,097 9,079 10,135Goodwill 117 304 304 304 304Other intangible assets 519 544 944 944 944Investments 5 81 81 81 81Other assets 392 574 10,574 10,574 10,574

Invested capital 3,949 4,184 19,000 20,982 22,038Cash & equivalents (558) (3,425) (2,497) (1,704) (1,985)Short term debt 870 1,773 1,773 1,773 1,773Long term debt * 328 574 4,774 4,974 5,174

Net debt 640 (1,078) 4,050 5,042 4,962Deferred tax 0 0 0 0 0Other liabilities 839 1,285 1,912 1,912 1,912Total equity 2,420 3,794 12,856 13,845 14,982Minority interests 50 182 182 182 182


Per share (RMB)

Book value per share 3.80 5.79 13.15 14.17 15.33Tangible book value per share 2.81 4.49 11.88 12.89 14.05

Financial strength

Net debt/equity (%) 25.9 (27.1) 31.1 35.9 32.7Net debt/total assets (%) 10.2 (9.5) 15.2 17.4 16.1Current ratio (x) 1.1 1.3 1.3 1.4 1.4

Valuation 2014A 2015A 2016E 2017E 2018E

Recurring P/E (x) * 61.9 55.3 47.7 29.2 24.2Recurring P/E @ target price (x) * 75.2 67.2 58.0 35.5 29.4Reported P/E (x) 61.9 55.3 47.7 29.2 24.2Dividend yield (%) 0.7 1.0 0.4 0.7 0.8P/CF (x) 26.5 21.8 32.0 22.3 19.2P/FCF (x) (1,036.9) (82.7) (21.6) (38.6) 107.6Price/book (x) 8.9 5.8 2.6 2.4 2.2Price/tangible book (x) 12.0 7.5 2.8 2.6 2.4EV/EBITDA (x) 27.3 22.9 23.9 19.1 16.2EV/EBITDA @ target price (x) 33.0 28.1 28.5 22.6 19.3


Source: Joyson Electronics, BNP Paribas estimates


103



Financial statementsPanasonic Corp

Profit and Loss (JPY b) Year Ending Mar 2015A 2016A 2017E 2018E 2019E

Revenue 7,715.0 7,553.7 7,418.4 7,663.0 8,218.0

Cost of sales (incl depreciation) (5,527.2) (5,340.0) (5,392.5) (5,523.6) (5,877.0)

Gross profit 2,187.8 2,213.7 2,025.9 2,139.4 2,341.0

SG&A (1,805.9) (1,798.0) (1,715.9) (1,774.4) (1,902.9)

R&D 0.0 0.0 0.0 0.0 0.0

Operating profit 381.9 415.7 310.0 365.0 438.1

Interest and dividends received 16.4 20.5 18.0 19.8 21.3

Associates 0.0 0.0 0.0 0.0 0.0

Other non-operating income 0.0 0.0 0.0 0.0 0.0

Interest paid (17.6) (17.0) (14.5) (14.5) (14.5)

Other non-operating expenses (198.3) (202.2) (9.5) (10.0) (9.9)

Extraordinary gains 0.0 0.0 0.0 0.0 0.0

Extraordinary losses 0.0 0.0 0.0 0.0 0.0

Pre-tax profit 182.5 217.0 306.5 360.3 434.9

Tax (2.0) (14.5) (104.1) (131.5) (143.5)

Minorities* (16.9) (21.8) (31.7) (32.6) (33.5)

Net profit 179.5 193.3 170.7 196.2 257.9

EBITDA (operating profit + depreciation) 668.4 690.5 600.0 656.2 758.6

Per share (JPY)

Reported EPS 77.6 83.4 73.6 84.6 111.1

DPS 18.0 25.0 30.0 32.0 34.0

Growth

Revenue (%) (0.3) (2.1) (1.8) 3.3 7.2

EBITDA (%) 14.5 3.3 (13.1) 9.4 15.6

Operating profit (%) 25.2 8.8 (25.4) 17.7 20.0

Recurring profit (%) - - - - -

EPS (%) 49.0 7.4 (11.8) 14.9 31.4



EBITDA margin (%) 8.7 9.1 8.1 8.6 9.2

Operating profit margin (%) 5.0 5.5 4.2 4.8 5.3

Net margin (%) 2.3 2.6 2.3 2.6 3.1


Dividend payout on net profit (%) 23.2 30.0 41.4 37.8 30.6

Interest cover (x) 1.9 2.1 51.5 76.8 139.0

Inventory days 50.0 51.9 50.7 49.9 49.4

Debtor days 47.3 43.9 40.2 39.2 38.5

Creditor days 63.4 65.8 63.2 62.2 61.5

Operating ROIC (%) 35.1 38.8 20.4 21.8 25.6

ROIC (%) 17.0 17.6 9.1 10.1 12.3

ROE (%) 10.6 11.0 9.5 10.4 12.7

ROA (%) 7.0 6.9 3.6 4.0 4.8

*Pre exceptional pre-goodwill and fully diluted

* Net profit for US GAAP. Source: Panasonic Corp, BNP Paribas estimates

104



Financial statementsPanasonic Corp

Cash Flow (JPY b) Year Ending Mar 2015A 2016A 2017E 2018E 2019EPre-tax profit* 182.5 217.0 306.5 360.3 434.9Tax paid 2.0 14.5 104.1 131.5 143.5Depreciation 286.5 274.8 290.0 291.2 320.5Working capital 0.0 0.0 0.0 0.0 0.0Other 20.5 (107.7) (196.7) (283.6) (333.9)Operating cash flow 491.5 398.7 503.9 499.3 565.0Capital expenditure (226.7) (248.8) (345.0) (344.8) (369.8)Other 88.7 (25.5) 0.0 0.0 0.0Investing cash flow (138.0) (274.3) (345.0) (344.8) (369.8)Free cash flow 353.5 124.4 158.9 154.5 195.2Change in debt 0.0 0.0 0.0 0.0 0.0Dividends paid (59.2) (64.4) (57.9) (69.6) (74.2)Net buybacks 0.0 0.0 0.0 0.0 0.0Equity issued 0.0 0.0 0.0 0.0 0.0Other 316.8 (243.6) 0.0 0.0 0.0Financing cash flow 257.6 (308.0) (57.9) (69.6) (74.2)Gross change in cash 687.9 (266.1) 100.9 84.9 121.0Other adjustments 18.5 (18.3) 0.0 0.0 0.0Change in cash 706.4 (284.5) 100.9 84.9 121.0

Per share (JPY)Operating cash flow per share 212.6 171.8 215.7 213.3 241.1FCF per share 152.9 53.7 68.5 66.6 84.1

Balance Sheet (JPY b) 2015A 2016A 2017E 2018E 2019ECash & equivalents 1,298.9 1,014.4 1,115.3 1,200.2 1,321.2A/c receivable 992.1 823.6 808.8 835.5 896.0Inventories 762.7 756.4 742.9 767.4 823.0Other 359.1 459.9 459.9 459.9 459.9Current assets 3,412.7 3,054.4 3,127.0 3,263.0 3,500.1Tangible fixed assets 1,374.8 1,301.2 1,356.2 1,409.8 1,459.1Intangible fixed assets 0.0 0.0 0.0 0.0 0.0Investments and other assets 1,169.4 1,241.4 1,241.4 1,241.4 1,241.4Total assets 5,956.9 5,597.0 5,724.6 5,914.3 6,200.7A/c payable 983.3 942.2 925.4 955.9 1,025.1Short term debt 260.5 21.7 21.7 21.7 21.7Other (1,489.0) (1,416.9) (1,416.9) (1,416.9) (1,416.9)Current liabilities 2,732.8 2,380.9 2,364.0 2,394.5 2,463.8Long term debt 712.4 704.2 704.2 704.2 704.2Other 519.2 657.6 657.6 657.6 657.6Long-term liabilities 1,231.6 1,361.8 1,361.8 1,361.8 1,361.8Total liabilities 3,964.4 3,742.7 3,725.8 3,756.3 3,825.5Common equity 1,823.3 1,705.1 1,817.9 1,944.5 2,128.1Preferred equity 0.0 0.0 0.0 0.0 0.0Minorities etc 169.3 149.3 181.0 213.5 247.1Net Assets 1,992.6 1,854.3 1,998.8 2,158.0 2,375.1Liabilities & net assets 5,956.9 5,597.0 5,724.6 5,914.3 6,200.7*includes convertibles and preferred stock which is being treated as debt

Per share (JPY)Book value per share 789 735 778 831 908Tangible book value per share 789 735 778 831 908

Financial strengthNet debt/equity (%) (16.4) (15.6) (19.5) (22.0) (25.1)Net debt/total assets (%) (5.5) (5.2) (6.8) (8.0) (9.6)Current ratio (x) 1.2 1.3 1.3 1.4 1.4CF interest cover (x) 2.8 1.6 27.4 33.5 63.0

Valuation 2015A 2016A 2017E 2018E 2019EP/E (x) * 13.4 12.4 14.3 12.3 9.3P/E @ target price (x) * 18.0 16.8 19.3 16.6 12.6Reported P/E (x) 13.4 12.4 14.1 12.3 9.3Dividend yield (%) 1.7 2.4 2.9 3.1 3.3P/CF (x) 4.9 6.0 4.8 4.8 4.3P/FCF (x) 6.8 19.3 15.2 15.6 12.3Price/book (x) 1.3 1.4 1.3 1.2 1.1Price/tangible book (x) 1.3 1.4 1.3 1.2 1.1EV/EBITDA (x) 3.4 3.3 3.7 3.3 2.7EV/EBITDA @ target price (x) 4.6 4.5 5.1 4.6 3.9EV/invested capital (x) 1.0 1.0 1.0 0.9 0.9

** Pre exceptional & pre-goodwill and fully diluted

* Net profit for US GAAP. Source: Panasonic Corp, BNP Paribas estimates

105



Financial statementsNidec Corp

Profit and Loss (JPY b) Year Ending Mar 2015A 2016A 2017E 2018E 2019E

Revenue 1,028.4 1,178.3 1,184.4 1,260.4 1,350.7

Cost of sales (incl depreciation) (786.2) (908.3) (911.5) (958.4) (1,022.5)

Gross profit 242.2 270.0 272.9 302.0 328.2

SG&A (131.2) (145.4) (142.9) (152.0) (162.9)

R&D 0.0 0.0 0.0 0.0 0.0

Operating profit 110.9 124.5 130.0 150.0 165.3

Interest and dividends received 2.4 1.9 2.4 2.5 2.9

Associates 0.0 0.0 0.0 0.0 0.0

Other non-operating income 0.8 (0.2) 0.0 0.0 0.0

Interest paid (1.5) (2.2) (2.4) (2.4) (2.4)

Other non-operating expenses (5.5) (4.7) 3.5 0.0 0.0

Extraordinary gains 0.0 0.0 0.0 0.0 0.0

Extraordinary losses 0.0 0.0 0.0 0.0 0.0

Pre-tax profit 107.1 119.3 127.6 150.0 165.7

Tax (29.1) (26.5) (30.0) (35.3) (38.9)

Minorities* (2.1) (1.1) (0.7) (0.8) (0.8)

Net profit 76.0 91.8 96.9 114.0 125.9

EBITDA (operating profit + depreciation) 162.4 189.2 192.0 221.6 241.5

Per share (JPY)

Reported EPS 256.1 308.2 326.7 384.3 424.5

DPS 70.0 80.0 100.0 120.0 130.0

Growth

Revenue (%) 17.5 14.6 0.5 6.4 7.2

EBITDA (%) 25.6 16.5 1.5 15.4 9.0

Operating profit (%) 30.7 12.3 4.4 15.4 10.2

Recurring profit (%) - - - - -

EPS (%) 32.0 20.4 6.0 17.6 10.4



EBITDA margin (%) 15.8 16.1 16.2 17.6 17.9

Operating profit margin (%) 10.8 10.6 11.0 11.9 12.2

Net margin (%) 7.4 7.8 8.7 9.1 9.3


Dividend payout on net profit (%) 27.3 26.0 28.9 31.2 30.6

Interest cover (x) 28.8 23.9 n/a n/a n/a

Inventory days 68.5 68.7 68.6 67.5 67.6

Debtor days 77.0 73.2 72.7 70.7 70.4

Creditor days 83.9 74.8 69.4 66.6 66.7

Operating ROIC (%) - - - - -

ROIC (%) - - - - -

ROE (%) 12.0 12.2 12.9 13.1 13.2

ROA (%) 6.4 7.1 7.1 7.7 8.0

*Pre exceptional pre-goodwill and fully diluted

Revenue By Division (JPY b) 2015A 2016A 2017E 2018E 2019E

HDD motor 204.1 208.0 166.2 156.6 145.5

Other small precision motors 193.9 240.0 242.4 261.0 278.3

ACI 263.0 283.4 303.3 330.6 360.3

Automotive 197.0 271.3 293.5 321.9 363.7

Machinery 98.8 106.5 110.6 117.8 126.1

Elec. and optical components 65.1 64.1 63.7 67.5 71.6

Other 6.5 5.0 4.8 5.0 5.2

* Net profit for US GAAP. Source: Nidec Corp, BNP Paribas estimates

106



Financial statementsNidec Corp

Cash Flow (JPY b) Year Ending Mar 2015A 2016A 2017E 2018E 2019EPre-tax profit* 107.1 119.3 127.6 150.0 165.7Tax paid 29.1 26.5 30.0 35.3 38.9Depreciation 51.4 64.7 62.0 71.6 76.2Working capital 0.0 0.0 0.0 0.0 0.0Other (36.5) (47.4) (60.0) (70.5) (77.9)Operating cash flow 91.9 147.6 149.7 171.0 184.7Capital expenditure (58.0) (81.9) (110.0) (90.0) (90.0)Other (23.2) (13.4) 0.0 0.0 0.0Investing cash flow (81.2) (95.3) (110.0) (90.0) (90.0)Free cash flow 10.6 52.3 39.7 81.0 94.7Change in debt 0.0 0.0 0.0 0.0 0.0Dividends paid (16.6) (24.0) (29.7) (35.6) (38.6)Net buybacks 0.0 0.0 0.0 0.0 0.0Equity issued 0.0 0.0 0.0 0.0 0.0Other (2.9) 31.8 0.0 0.0 0.0Financing cash flow (19.5) 7.8 (29.7) (35.6) (38.6)Gross change in cash 22.2 36.0 10.1 45.4 56.2Other adjustments 0.0 0.0 0.0 0.0 0.0Change in cash 22.2 36.0 10.1 45.4 56.2

Per share (JPY)Operating cash flow per share 312.5 497.7 504.8 576.6 622.8FCF per share 35.9 175.5 133.9 273.2 319.3

Balance Sheet (JPY b) 2015A 2016A 2017E 2018E 2019ECash & equivalents 269.9 305.9 316.0 361.4 417.6A/c receivable 237.6 235.3 236.5 251.7 269.7Inventories 170.9 171.0 171.8 182.9 196.0Other 50.6 53.2 53.2 53.2 53.2Current assets 729.0 765.3 777.5 849.1 936.4Tangible fixed assets 342.1 347.7 395.7 414.1 427.9Intangible fixed assets 163.0 163.0 163.0 163.0 163.0Investments and other assets 123.2 108.5 108.5 108.5 108.5Total assets 1,357.3 1,384.5 1,444.6 1,534.7 1,635.8A/c payable 195.0 177.3 169.5 180.3 193.3Short term debt 97.9 163.9 163.9 163.9 163.9Other (70.1) (79.3) (79.3) (79.3) (79.3)Current liabilities 362.9 420.5 412.7 423.6 436.5Long term debt 184.6 136.9 136.9 136.9 136.9Other 56.7 54.6 54.6 54.6 54.6Long-term liabilities 241.3 191.5 191.5 191.5 191.5Total liabilities 604.2 612.0 604.2 615.1 628.0Common equity 745.0 764.2 831.5 909.9 997.2Preferred equity 0.0 0.0 0.0 0.0 0.0Minorities etc 8.1 8.3 9.0 9.7 10.6Net Assets 753.1 772.5 840.5 919.6 1,007.8Liabilities & net assets 1,357.3 1,384.5 1,444.6 1,534.7 1,635.8*includes convertibles and preferred stock which is being treated as debt

Per share (JPY)Book value per share 2,534 2,577 2,803 3,068 3,362Tangible book value per share 1,979 2,027 2,254 2,518 2,813

Financial strengthNet debt/equity (%) 1.7 (0.7) (1.8) (6.6) (11.6)Net debt/total assets (%) 0.9 (0.4) (1.1) (4.0) (7.1)Current ratio (x) 2.0 1.8 1.9 2.0 2.1CF interest cover (x) 3.8 11.0 n/a n/a n/a

Valuation 2015A 2016A 2017E 2018E 2019EP/E (x) * 36.4 30.3 26.9 24.2 21.9P/E @ target price (x) * 41.8 34.7 30.9 27.8 25.2Reported P/E (x) 36.4 30.3 28.5 24.3 22.0Dividend yield (%) 0.8 0.9 1.1 1.3 1.4P/CF (x) 18.3 17.0 17.3 14.8 13.6P/FCF (x) 260.0 53.1 69.6 34.1 29.2Price/book (x) 3.7 3.6 3.3 3.0 2.8Price/tangible book (x) 4.7 4.6 4.1 3.7 3.3EV/EBITDA (x) 17.0 14.6 14.4 12.2 11.0EV/EBITDA @ target price (x) 19.5 16.8 16.5 14.1 12.7EV/invested capital (x) 3.4 3.4 3.1 3.0 2.8

** Pre exceptional & pre-goodwill and fully diluted

* Net profit for US GAAP. Source: Nidec Corp, BNP Paribas estimates

107



Disclaimers and Disclosures

APPENDIX

DISCLAIMERS AND DISCLOSURES APPLICABLE TO NON-US BROKER-DEALER(S): BNP PARIBAS SECURITIES (ASIA) LTD, BNP PARIBAS SECURITIES (JAPAN) LTD

ANALYST(S) CERTIFICATION

Jimmy Huang, BNP Paribas Securities (Asia) Ltd, +852 2825 1124, [email protected] Vey-Sern Ling, BNP Paribas Securities (Asia) Ltd, +852 2825 1802, [email protected] Laura Chen, BNP Paribas Securities (Asia) Ltd, +852 2825 1280, [email protected] Masahiro Wakasugi, BNP Paribas Securities (Japan) Ltd, +813 6377 2240, [email protected] The BNP Paribas Securities (Asia) Ltd, BNP Paribas Securities (Japan) Ltd Analysts mentioned in this disclaimer are employed by a non-US affiliate of BNP Paribas Securities Corp., and are not registered/ qualified pursuant to NYSE and/or FINRA regulations

The individual(s) identified above certify(ies) that (i) all views expressed in this report accurately reflect the personal view of the analyst(s) with regard to any and all of the subject securities, companies or issuers mentioned in this report; and (ii) no part of the compensation of the analyst(s) was, is, or will be, directly or indirectly, related to the specific recommendations or views expressed herein.

IMPORTANT DISCLOSURES REQUIRED IN THE UNITED STATES BY FINRA RULES AND OTHER JURISDICTIONS "BNP Paribas” is the marketing name for the global banking and markets business of BNP Paribas Group. No portion of this report was prepared by BNP Paribas Securities Corp (US) personnel, and it is considered Third-Party Affiliate research under NASD Rule 2711. The following disclosures relate to relationships between companies covered in this research report and the BNP entity identified on the cover of this report, BNP Securities Corp., and other entities within the BNP Paribas Group (collectively, "BNP Paribas"). The disclosure column in the following table lists the important disclosures applicable to each company that has been rated and/or recommended in this report:

BNP Paribas represents that: 1. Within the past year, it has managed or co-managed a public offering for this company, for which it received fees. 2. It had an investment banking relationship with this company in the last 12 months. 3. It received compensation for investment banking services from this company in the last 12 months. 4. It expects to receive or intends to seek compensation for investment banking services from the subject company/ies in the next 3 months. 5. It beneficially owns 1% or more of any class of common equity securities of the subject company. 6. It makes a market in securities in respect of this company. 7. The analyst(s) or an individual who assisted in the preparation of this report (or a member of his/her household) has a financial interest position in

securities issued by this company. The financial interest is in the common stock of the subject company, unless otherwise noted. 8. The analyst (or a member of his/her household) is an officer, director, employee or advisory board member of this company or has received

compensation from the company.

IMPORTANT DISCLOSURES REQUIRED IN KOREA The disclosure column in the following table lists the important disclosures applicable to each Korea listed company that has been rated and/or recommended in this report:

1. The performance of obligations of the Company is directly or indirectly guaranteed by BNP Paribas Securities Korea Co. Ltd (“BNPPSK”) by means of payment guarantees, endorsements, and provision of collaterals and/or taking over the obligations.

2. BNPPSK owns 1/100 or more of the total outstanding shares issued by the Company. 3. The Company is an affiliate of BNPPSK as prescribed by Item 3, Article 2 of the Monopoly Regulation and Fair Trade Act. 4. BNPPSK is the financial advisory agent of the Company for the Merger and Acquisition transaction or of the Target Company whereby the size of the

transaction does not exceed 5/100 of the total asset of the Company or the total number of outstanding shares. 5. BNPPSK has taken financial advisory service regarding listing to the Company within the past 1 year. 6. With regards to the tender offer initiated by the Company based on Item 2, Article 133 of the Financial Investment Services and Capital Market Act,

BNPPSK acts in the capacity of the agent for the tender offer designated either by the Company or by the target company, provided that this provision shall apply only where tender offer has not expired.

7. The listed company which issued the stocks in question in case where 40 days has not passed since the new shares were listed from the date of entering into arrangement for public offering or underwriting-related agreement for issuance of stocks

8. The Company that has signed a nominated advisor contract with BNPPSK as defined in Item 2 of Article 8 of the KONEX Market Listing Regulation. 9. The Company is recognized as having considerable interests with BNPPSK in relation to No.1 to No. 8. 10. The analyst or his/her spouse owns (including delivery claims of marketable securities based on legal regulations and trading and misc. contracts) the

following securities or rights (hereinafter referred to as “Securities, etc.” in this Article) regardless of whose name is used in the trading. 1) Stocks, bond with stock certificate, and certificate of pre-emptive rights issued by the Company whose securities dealings are being solicited. 2) Stock options of the Company whose securities dealings are being solicited. 3) Individual stock future, stock option, and warrants that use the stocks specified in Item 1) as underlying.

Company Ticker Disclosure (as applicable)

Panasonic Corp 6752 JP 2, 3, 4

Wistron NeWeb 6285 TT 2, 3, 4

Company Ticker Price (as of 29-Sep-2016 closing price) InterestN/A N/A N/A N/A

108



Company Ticker Price Rating Valuation & Risks

Baidu BIDU US USD 182.23 Buy Key risks to our DCF-based TP include: 1) aggressive competition in mobile search impacting Baidu's traffic share and hampering monetisation ramp-up, leading to lower-than-expected revenue growth, 2) taking longer than expected to secure mobile dominance and mobile user stickiness, leading to sustained marketing expenses and weaker margin recovery and 3) continued high level of investment in O2O initiatives without clear returns on investment.

Iflytek 002230 CH RMB 28.43 Hold Downside risks to our P/E-based TP are 1) worse-than-expected China voice recognition demand, 2) international/domestic players penetrating the China voice recognition market, and 3) worse-than-expected growth of China smart car market Upside risks are 1) better-than-expected China voice recognition demand, 2) government policy support to China local voice recognition players, and 3) better-than-expected growth of China smart car market

Joyson Electronics 600699 CH RMB 33.73 Buy Downside risk to our P/E-based TP include: 1) softer global auto market, 2) currency trends, especially EUR depreciation, 3) market competition, 4) market share loss to international competitors, 5) difficulties integrating overseas and domestic business units, and 6) slower-than-expected EV development in China and overseas.

Nidec Corp 6594 JP JPY 9,323 Buy Our TP of JPY10,700 is based on our FY3/18E P/E of 27.6x. Risks include: 1) delays in regulations on automotive, appliance, commercial and industrial motors; 2) slowing volumes and rising price competition due to a slump in demand; 3) a greater-than-forecast decline in HDD volumes; and 4) rapid JPY appreciation.

Panasonic Corp 6752 JP JPY 1,038 Buy Our TP of JPY1,400 is based on a P/E of 16x FY3/18E. Downside risks include 1) Slowdown in auto demand, 2) delays in restructuring of TV and semiconductor businesses, and 3) steep emerging market currency depreciation should be negative for earnings.

Tung Thih Electronic 3552 TT TWD 448.00 Buy Our valuation is based on DCF. Downside risks include: 1) greater-than-expected China auto market growth slowdown, especially SUVs, amid the slowdown in China’s economy, 2) slower-than-expected adoption of D2xx models, and 3) FX volatility, especially RMB.

Wistron NeWeb 6285 TT TWD 98.00 Buy Key downside risks to our P/E-based TP are weaker-than-expected automobile sales, slower-than-expected ADAS adoption and unexpected changes in regulations that reinforce the adoption of ADAS.

Sources: Factset, BNP Paribas

GENERAL DISCLAIMER

This report was produced by BNP Paribas Securities (Asia) Ltd, BNP Paribas Securities (Japan) Ltd, member company(ies) of the BNP Paribas Group.

This report is for the use of intended recipients only and may not be reproduced (in whole or in part) or delivered or transmitted to any other person without our prior written consent. By accepting this report, the recipient agrees to be bound by the terms and limitations set forth herein. This report does not constitute a personal recommendation or take into account the particular investment objectives, financial situations, or needs of individual clients. Customers are advised to use the information contained herein as just one of many inputs and considerations prior to engaging in any trading activity. This report does not constitute a prospectus or other offering document or an offer or solicitation to buy or sell any securities or other investments. This report is not intended to provide the sole basis of any evaluation of the subject securities and companies mentioned in this report. Information and opinions contained in this report are published for reference of the recipients and are not to be relied upon as authoritative or without the recipient’s own independent verification, or taken in substitution for the exercise of judgment by the recipient. Additionally, the products mentioned in this report may not be available for sale in certain jurisdictions. As an investment bank with a wide range of activities, BNP Paribas may face conflicts of interest, which are resolved under applicable legal provisions and internal guidelines. You should be aware, however, that BNP Paribas may engage in transactions in a manner inconsistent with the views expressed in this document, either for its own account or for the account of its clients. Australia: This report is being distributed in Australia by BNP Paribas Sydney Branch, registered in Australia as ABN 23 000 000 117 at 60 Castlereagh Street Sydney NSW 2000. BNP Paribas Sydney Branch is licensed under the Banking Act 1959 and the holder of Australian Financial Services Licence no. 238043 and therefore subject to regulation by the Australian Securities & Investments Commission in relation to delivery of financial services. By accepting this document you agree to be bound by the foregoing limitations, and acknowledge that information and opinions in this document relate to financial products or financial services which are delivered solely to wholesale clients (in terms of the Corporations Act 2001, sections 761G and 761GA; Corporations Regulations 2001, division 2, reg. 7.1.18 & 7.1.19) and/or professional investors (as defined in section 9 of the Corporations Act 2001). Canada: The information contained herein is not, and under no circumstances is to be construed as, a prospectus, an advertisement, a public offering, an offer to sell securities described herein, or solicitation of an offer to buy securities described herein, in Canada or any province or territory thereof. Any offer or sale of the securities described herein in Canada will be made only under an exemption from the requirements to file a prospectus with the relevant Canadian securities regulators and only by a dealer properly registered under applicable securities laws or, alternatively, pursuant to an exemption from the dealer registration requirement in the relevant province or territory of Canada in which such offer or sale is made. The information contained herein is under no circumstances to be construed as investment advice in any province or territory of Canada and is not tailored to the needs of the recipient. To the extent that the information contained herein references securities of an issuer incorporated, formed or created under the laws of Canada or a province or territory of Canada, any trades in such securities must be conducted through a dealer registered in Canada. No securities commission or similar regulatory authority in Canada has reviewed or in any way passed judgment upon these materials, the information contained herein or the merits of the securities described herein, and any representation to the contrary is an offence. Hong Kong: This report is prepared for professional investors and is being distributed in Hong Kong by BNP Paribas Securities (Asia) Limited to persons whose business involves the acquisition, disposal or holding of securities, whether as principal or agent. BNP Paribas Securities (Asia) Limited, a subsidiary of BNP Paribas, is regulated by the Securities and Futures Commission for the conduct of dealing in securities, advising on securities, dealing in futures contracts and advising on corporate finance. For professional investors in Hong Kong, please contact BNP Paribas Securities (Asia) Limited (address: 63/F Two International Finance Centre, 8 Finance Street, Central, Hong Kong; tel:2909 8888; fax: 2845 2232) for all matters and queries relating to this report. India: In India, this document is being distributed by BNP Paribas Securities India Pvt. Ltd. ("BNPPSIPL"), having its registered office at 5th floor, BNP Paribas House, 1 North Avenue, Maker Maxity, Bandra Kurla Complex, Bandra (East), Mumbai 400 051, INDIA (Tel. no. +91 22 3370 4000 / 6196 4000, Fax no. +91 22 6196 4363). BNPPSIPL is registered with the Securities and Exchange Board of India (“SEBI”) as a research analyst (Regn. No. INH000000792) and as a stockbroker in the Equities and the Futures & Options segments of National Stock Exchange of India Ltd. (“NSE”) and BSE Ltd. and in the Currency Derivatives segment of NSE (SEBI Regn. Nos.: INB/INF/NSF/NSE231474835, INB/INF011474831; CIN: U74920MH2008FTC182807; Website: www.bnpparibas.co.in). No material disciplinary action has been taken against BNPPSIPL by any regulatory or government authority. BNPPSIPL or its associates may have received compensation or other benefits for brokerage services or for other products or services, from the company(ies) that have been rated and/or recommended in the report and / or from third parties. Indonesia: This report is being distributed by PT BNP Paribas Securities Indonesia and is delivered by its licensed employee(s), including marketing/sales person, to its client. PT BNP Paribas Securities Indonesia, having its registered office at Menara BCA, 35th floor, Grand Indonesia, JL. M.H. Thamrin No.1, Jakarta 10310, Indonesia, is a subsidiary company of BNP Paribas SA and licensed under Capital Market Law no. 8 year 1995, a holder of broker-dealer and underwriter licenses issued by the Capital Market and Financial Institution Supervisory Agency (now Otoritas Jasa Keuangan/OJK). PT BNP Paribas Securities Indonesia is also a member of Indonesia Stock Exchange and supervised by Otoritas Jasa Keuangan (OJK). Neither this report nor any copy hereof may be distributed in Indonesia or to any Indonesian citizens except in compliance with applicable Indonesian capital market laws and regulations. This report is not an offer of securities in Indonesia and may not be distributed within the territory of the Republic of Indonesia or to Indonesian citizens in circumstance which constitutes an offering within the meaning of Indonesian capital market laws and regulations. Japan: This report is being distributed to Japanese based firms by BNP Paribas Securities (Japan) Limited or by a subsidiary or affiliate of BNP Paribas not

109



registered as a financial instruments firm in Japan, to certain financial institutions defined by article 17-3, item 1 of the Financial Instruments and Exchange Law Enforcement Order. BNP Paribas Securities (Japan) Limited is a financial instruments firm registered according to the Financial Instruments and Exchange Law of Japan and a member of the Japan Securities Dealers Association, the Financial Futures Association of Japan and the Type II Financial Instruments Firms Association. BNP Paribas Securities (Japan) Limited accepts responsibility for the content of a report prepared by another non-Japan affiliate only when distributed to Japanese based firms by BNP Paribas Securities (Japan) Limited. Some of the foreign securities stated on this report are not disclosed according to the Financial Instruments and Exchange Law of Japan. Malaysia: This report is issued and distributed by BNP Paribas Capital (Malaysia) Sdn Bhd. The views and opinions in this research report are our own as of the date hereof and are subject to change. BNP Paribas Capital (Malaysia) Sdn Bhd has no obligation to update its opinion or the information in this research report. This publication is strictly confidential and is for private circulation only to clients of BNP Paribas Capital (Malaysia) Sdn Bhd. This publication is being provided to you strictly on the basis that it will remain confidential. No part of this material may be (i) copied, photocopied, duplicated, stored or reproduced in any form by any means or (ii) redistributed or passed on, directly or indirectly, to any other person in whole or in part, for any purpose without the prior written consent of BNP Paribas Capital (Malaysia) Sdn Bhd. Philippines: This report is being distributed in the Philippines by BNP Paribas Manila Branch, an Offshore Banking Unit (OBU) of BNP Paribas whose head office is in Paris, France. BNP Paribas Manila OBU is registered as an offshore banking unit under Presidential Decree No. 1034 (PD 1034), and regulated by the Bangko Sentral ng Pilipinas. This report is being distributed in the Philippines to qualified clients of OBUs as allowed under PD 1034, and is qualified in its entirety to the products and services allowed under PD 1034. Singapore: This report is distributed in Singapore by BNP Paribas Securities (Singapore) Pte Ltd ("BNPPSSL") and may be distributed in Singapore only to an Accredited or Institutional Investor, each as defined under the Financial Advisers Regulations ("FAR") and the Securities and Futures Act (Chapter 289) of Singapore, as amended from time to time. In relation to the distribution to such categories of investors, BNPPSSL and its representatives are exempted under Regulation 35 of the FAR from the requirements in Section 36 of the Financial Advisers Act of Singapore, regarding the disclosure of certain interests in, or certain interests in the acquisition or disposal of, securities referred to in this report. For Institutional and Accredited Investors in Singapore, please contact BNP Paribas Securities (Singapore) Ptd Ltd (company registration number: 199801966C; address: 10 Collyer Quay, 34/F Ocean Financial Centre, Singapore 049315; tel: (65) 6210 1288; fax: (65) 6210 1980) for all matters and queries relating to this report. South Africa: In South Africa, BNP Paribas Securities South Africa (Pty) Ltd is a licensed member of the Johannesburg Stock Exchange and an authorised Financial Services Providers and subject to regulation by the Financial Services Board. BNP Paribas Securities South Africa (Pty) Ltd does not expressly or by implication represent, recommend or propose that the financial products referred to in this report are appropriate to the particular investment objectives, financial situation or particular needs of the recipient. This document does not constitute advice as contemplated in the Financial Advisory and Intermediary Services Act, 2002. South Korea: BNP Paribas Securities Korea is registered as a Licensed Financial Investment Business Entity under the FINANCIAL INVESTMENT SERVICES AND CAPITAL MARKETS ACT and regulated by the Financial Supervisory Service and Financial Services Commission. This document does not constitute an offer to sell to or the solicitation of an offer to buy from any person any financial products where it is unlawful to make the offer or solicitation in South Korea. Switzerland: This report is intended solely for customers who are “Qualified Investors” as defined in article 10 paragraphs 3 and 4 of the Swiss Federal Act on Collective Investment Schemes of 23 June 2006 (CISA) and the relevant provisions of the Swiss Federal Ordinance on Collective Investment Schemes of 22 November 2006 (CISO). “Qualified Investors” includes, among others, regulated financial intermediaries such as banks, securities dealers, fund management companies and asset managers of collective investment schemes, regulated insurance companies as well as pension funds and companies with professional treasury operations. This document may not be suitable for customers who are not Qualified Investors and should only be used and passed on to Qualified Investors. For specification purposes, a “Swiss Corporate Customer” is a Client which is a corporate entity, incorporated and existing under the laws of Switzerland and which qualifies as “Qualified Investor” as defined above." BNP Paribas (Suisse) SA is authorised as bank and as securities dealer by the Swiss Federal Market Supervisory Authority FINMA. BNP Paribas (Suisse) SA is registered at the Geneva commercial register under No. CH-270-3000542-1. BNP Paribas (Suisse) SA is incorporated in Switzerland with limited liability. Registered Office: 2 place de Hollande, CH-1204 Geneva. Taiwan: This report is being distributed to Taiwan based clients by BNP Paribas Securities (Taiwan) Co., Ltd or by a subsidiary or affiliate of BNP Paribas. Such information is for your reference only. The reader should independently evaluate the investment risks and is solely responsible for their investment decision. Information on securities that do not trade in Taiwan is for informational purposes only and is not to be construed as a recommendation or a solicitation to trade in such securities. BNP Paribas Securities (Taiwan) Co., Ltd. may not execute transactions for clients in these securities. This publication may not be distributed to the public media or quoted or used by the public media without the express written consent of BNP Paribas.

Thailand: Research relating to Thailand and Thailand based issuers is produced pursuant to an arrangement between BNP PARIBAS (“BNPP”) and Finansia Syrus Securities Public Company Limited (“FSS”). FSS International Investment Advisory Securities Co Ltd (“FSSIA”) prepares and distributes research under the brand name “BNP PARIBAS/FSS”. BNPP is not an affiliate of FSSIA or FSS. FSS also publishes a different research product under the brand name “FINANSIA SYRUS,” which is prepared by research analysts who are not part of FSSIA and who may cover the same securities, issuers, or industries that are the subject of this report. The ratings, recommendations, and views expressed in this report may differ from the ratings, recommendations, and views expressed by other research analysts or research teams employed by FSS. This report is being distributed outside Thailand by members of BNP Paribas. Turkey: This report is being distributed in Turkey by TEB Investment (TEB YATIRIM MENKUL DEGERLER A.S., Teb Kampus D Blok Saray Mah. Kucuksu Cad. Sokullu Sok., No:7 34768 Umraniye, Istanbul, Turkey, Trade register number: 358354, www.tebyatirim.com.tr) and outside Turkey jointly by TEB Investment and BNP Paribas. Information, comments and suggestions on investment given in this material are not within the scope of investment consulting. The investment consulting services are rendered tailor made for individuals by competent authorities considering the individuals’ risk and return preferences. However the comments and recommendations herein are based on general principles. These opinions may not be consistent with your financial status as well as your risk and return preferences. Therefore, making an investment decision only based on the information provided herein may not bear consequences in parallel with your expectations. This material issued by TEB Yatırım Menkul Değerler A.Ş. for information purposes only and may be changed without any prior notification. All rights reserved. No part of this material may be copied or reproduced in any manner without the written consent of TEB Yatırım Menkul Değerler A.Ş. Although TEB Yatırım Menkul Değerler A.Ş. gathers the presented material that is current as possible, it does not undertake that all the information is accurate or complete, nor should it be relied upon as such. TEB Yatırım Menkul Değerler A.Ş. assumes no responsibility whatsoever in respect of or arising out or in connection with the content of this material to third parties. If any third party chooses to use the content of this material as reference, he/she accepts and approves to do so entirely at his/her own risk.

United States: This report may be distributed in the United States only to U.S. Persons who are “major U.S. institutional investors” (as such term is defined in Rule 15a-6 under the Securities Exchange Act of 1934, as amended) and is not intended for the use of any person or entity that is not a “major U.S. institutional investor”. U.S persons who wish to effect transactions in securities discussed herein must do so through BNP Paribas Securities Corp., a US-registered broker dealer and member of FINRA, SIPC, NFA, NYSE and other principal exchanges. Certain countries within the European Economic Area: This document may only be distributed in the United Kingdom to eligible counterparties and professional clients and is not intended for, and should not be circulated to, retail clients (as such terms are defined in the Markets in Financial Instruments Directive 2004/39/EC (“MiFID”)). This document will have been approved for publication and distribution in the United Kingdom by BNP Paribas London Branch, a branch of BNP Paribas SA whose head office is in Paris, France. BNP Paribas SA is incorporated in France with limited liability with its registered office at 16 boulevard des Italiens, 75009 Paris. BNP Paribas London Branch (registered office: 10 Harewood Avenue, London NW1 6AA; tel: [44 20] 7595 2000; fax: [44 20] 7595 2555) is lead supervised by the European Central Bank (ECB) and the Autorité de Contrôle Prudentiel et de Résolution (ACPR). BNP Paribas London Branch is authorised by the ACPR and the Prudential Regulation Authority (PRA) and subject to limited regulation by the Financial Conduct Authority and PRA. Details about the extent of our authorisation and regulation by the PRA, and regulation by the Financial Conduct Authority are available from us on request. This report has been approved for publication in France by BNP Paribas, a credit institution licensed as an investment services provider by the ACPR whose head office is 16, Boulevard des Italiens 75009 Paris, France. This report is being distributed in Germany either by BNP Paribas London Branch or by BNP Paribas Niederlassung Frankfurt am Main, regulated by the Bundesanstalt für Finanzdienstleistungsaufsicht (BaFin). Other Jurisdictions: The distribution of this report in other jurisdictions or to residents of other jurisdictions may also be restricted by law, and persons into whose possession this report comes should inform themselves about, and observe, any such restrictions. By accepting this report you agree to be bound by

110



the foregoing instructions. This report is not directed to, or intended for distribution to or use by, any person or entity that is a citizen or resident of or located in any locality, state, country, or other jurisdiction where such distribution, publication, availability or use would be contrary to law or regulation. All research reports are disseminated and available to all clients simultaneously through our internal client websites. For all research available on a particular stock, please contact the relevant BNP Paribas research team or the author(s) of this report.

Additional Disclosures Target price history, stock price charts, valuation and risk details, and equity rating histories applicable to each company rated in this report is available in our most recently published reports available on our website: http://eqresearch.bnpparibas.com, or you can contact the analyst named on the front of this note or your BNP Paribas representative. BNP Paribas Personal Finance and Geely Automobile Holdings Limited have entered into an agreement in December 2013 to establish a joint venture company to engage in auto financing in the People’s Republic of China. All share prices are as at market close on 29 September 2016 unless otherwise stated.

RECOMMENDATION STRUCTURE

Stock Ratings Stock ratings are based on absolute upside or downside, which we define as (target price* - current price) / current price. BUY (B). The upside is 10% or more. HOLD (H). The upside or downside is less than 10%. REDUCE (R). The downside is 10% or more. Unless otherwise specified, these recommendations are set with a 12-month horizon. Thus, it is possible that future price volatility may cause a temporary mismatch between upside/downside for a stock based on market price and the formal recommendation. * In most cases, the target price will equal the analyst's assessment of the current fair value of the stock. However, if the analyst doesn't think the market will reassess the stock over the specified time horizon due to a lack of events or catalysts, then the target price may differ from fair value. In most cases, therefore, our recommendation is an assessment of the mismatch between current market price and our assessment of current fair value. Industry Recommendations Improving (): The analyst expects the fundamental conditions of the sector to be positive over the next 12 months. Stable (previously known as Neutral) (): The analyst expects the fundamental conditions of the sector to be maintained over the next 12 months. Deteriorating (): The analyst expects the fundamental conditions of the sector to be negative over the next 12 months. Country (Strategy) Recommendations Overweight (O). Over the next 12 months, the analyst expects the market to score positively on two or more of the criteria used to determine market recommendations: index returns relative to the regional benchmark, index sharpe ratio relative to the regional benchmark and index returns relative to the market cost of equity. Neutral (N). Over the next 12 months, the analyst expects the market to score positively on one of the criteria used to determine market recommendations: index returns relative to the regional benchmark, index sharpe ratio relative to the regional benchmark and index returns relative to the market cost of equity. Underweight (U). Over the next 12 months, the analyst does not expect the market to score positively on any of the criteria used to determine market recommendations: index returns relative to the regional benchmark, index sharpe ratio relative to the regional benchmark and index returns relative to the market cost of equity.

RATING DISTRIBUTION (as at 30 September 2016)

Should you require additional information concerning this report please contact the relevant BNP Paribas research team or the author(s) of this report. © 2016 BNP Paribas Group

Total BNP Paribas coverage universe 456 Investment Banking Relationship (%)

Buy 266 (58.3%) Buy 31.20

Hold 132 (28.9%) Hold 36.36

Reduce 58 (12.7%) Reduce 25.86

111

.

.

HONG KONG BNP Paribas Securities (Asia) Ltd 63/F, Two International Finance Centre 8 Finance Street, Central Hong Kong SAR China Tel (852) 2825 1888 Fax (852) 2845 9411

SHANGHAI

KUALA LUMPUR

Malaysia

NEW YORK BNP Paribas The Equitable Tower 787 Seventh Avenue New York NY 10019, USA Tel (1 212) 841 3800 Fax (1 212) 841 3810

BASEL BNP Paribas Aeschengraben 26 CH 4002 Basel Switzerland Tel (41 61) 276 5555 Fax (41 61) 276 5514

FRANKFURT BNP Paribas Mainzer Landstrasse 16 60325 Frankfurt Germany Tel (49 69) 7193 6637 Fax (49 69) 7193 2520

GENEVA BNP Paribas 2 Place de Hollande 1211 Geneva 11 Switzerland Tel (41 22) 787 7377 Fax (41 22) 787 8020

LONDON BNP Paribas 10 Harewood Avenue London NW1 6AA UK Tel (44 20) 7595 2000 Fax (44 20) 7595 2555

MADRID BNP Paribas SA, sucursal en Espana Hermanos Becquer 3 PO Box 50784 28006 Madrid Spain Tel (34 91) 745 9000 Fax (34 91) 745 8888

MILAN BNP Paribas Equities Italia SIM SpA Piazza San Fedele, 2 20121 Milan Italy Tel (39 02) 72 47 1 Fax (39 02) 72 47 6562

PARIS BNP Paribas Equities France Société de Bourse 20 boulevard des Italiens 75009 Paris France Tel (33 1) 4014 9673 Fax (33 1) 4014 0066

ZURICH BNP Paribas Talstrasse 41 8022 Zurich Switzerland Tel (41 1) 229 6891 Fax (41 1) 267 6813

MANAMA BNP Paribas Bahrain PO Box 5253 Manama Bahrain Tel (973) 53 3978 Fax (973) 53 1237

TOKYO BNP Paribas Securities (Japan) Ltd GranTokyo North Tower 1-9-1 Marunouchi, Chiyoda-Ku Tokyo 100-6740 Japan Tel (81 3) 6377 2000 Fax (81 3) 5218 5970

MUMBAI

BNP Paribas Equities (Asia) Ltd Shanghai Representative Office Room 2630, 26/F Shanghai World Financial Center 100 Century Avenue Shanghai 200120, China Tel (86 21) 6096 9000 Fax (86 21) 6096 9018

JAKARTA PT BNP Paribas Securities Indonesia Grand Indonesia, Menara BCA, JI. M.H. Thamrin No. 1Jakarta 10 0 Indonesia Tel (62 21) 2358 6586 Fax (62 21) 2358 7587

35/F

31

TAIPEI BNP Paribas Securities (Taiwan) Co Ltd 72 F, Taipei 101 No. 7 Xin Yi Road, Sec. 5 Taipei, Taiwan

(886 2) 8729 7000 Fax (886 2) 8101 2168

Tel

/Vista Tower, Level 48CThe Intermark, 182 Jalan Tun Razak50400 Kuala Lumpur

Tel (60 3) 2179 6222Fax (60 3) 2179 6226

BNP Paribas Capital (Malaysia) Sdn Bhd BNP Paribas Securities India Pvt Ltd BNP Paribas House1 North Avenue, Maker MaxityBandra Kurla ComplexBandra EastMumbai 400 051Tel (91 22) 3370 4000Fax (91 22) 3370 4386

https://eqresearch.bnpparibas.com

TEB Investment (A JV between TEB Bank and BNP Paribas) TEB Kampus D7 Saray Mahallesi Sokullu Sok No 7 Umraniye 34768 Istanbul Turkey Tel: (90 216) 636 44 44 Fax: (90 216) 631 44 00

SINGAPORE BNP Paribas Securities (Singapore) Pte Ltd (Co. Reg. No. 199801966C)

Tel (65) 6210 1288 Fax (65) 6210 1980

10 Collyer Quay

Singapore 049315 34/F Ocean Financial Centre

ISTANBUL CAPE TOWN Ground floor, Fernwood House The Oval, 1 Oakdale Road, Newlands Cape Town South Africa 7700 Tel (27 21) 657 8300 Fax (27 21) 657 8301

BEIJING BNP Paribas (China) Ltd Beijing Branch Room 2001, 20/F China World Tower 1 Jianguomenwai Avenue Beijing, China Tel: +86-10-6535 0888 Fax: +86-10-6535 0883

BANGKOK(In cooperation with BNP Paribas) FSS International Investment Advisory Securities Co., Ltd 990 Abdulrahim Place, 12/F, Room 1210 Rama IV Road, Bangrak Bangkok 10500 Thailand Tel (66 2) 611 3500 Fax (66 2) 611 3551

SEOULBNP Paribas Securities Korea Co Ltd 25/F, State Tower Namsan 100 Toegye-Ro Jung-Gu, Seoul 100-052 Tel (82 2) 2125 0500 Fax (82 2) 2125 0593

BNP Paribas Securities South Africa (Pty) Ltd

SECTOR REPORT ASIA TECHNOLOGY - BNP Paribas · 2016-09-30 · Autonomous driving is a goal for...

Documents

Transcript of SECTOR REPORT ASIA TECHNOLOGY - BNP Paribas · 2016-09-30 · Autonomous driving is a goal for...