Why is Mobileye falling behind?


 Recently, Mobileye, the autonomous driving division of Intel (Intel), once again submitted a prospectus to Nasdaq, with the stock code "MBLY".


  This is Mobileye's second IPO, but its valuation has plummeted. Mobileye was valued at about $50 billion when Intel "secretly" filed with the SEC in March. However, Mobileye has recently updated its prospectus again. Based on the number of shares issued and the issue price, its valuation has been reduced to about US$14.4-16 billion, which is similar to the valuation at the time of Intel's acquisition.


  Although the decline in valuation cannot ignore macro factors, more importantly, the overlord of the ADAS era is gradually losing its edge.


On the one hand, in the face of the siege of startup competitors such as NVIDIA, Qualcomm, and even Horizon, Mobileye's computing power is stretched. Compared with the 254TOPS computing power of NVIDIA's Orin chip in the same period, the computing power of EYEQ5 is only 24TOPS; on the other hand, The "black box" solution with highly coupled software and hardware has been unable to adapt to the development of car companies. New car companies such as Weilai and Ideal have abandoned them one after another, and even traditional car companies such as BMW, Volkswagen, GM, Mercedes-Benz and other customers have followed suit. Switch to another platform.


  In the new round of competition in the ADAS market, Mobileye is slowly falling behind.


  But even if the macro environment is not optimistic and the business is cold, Intel still wants to list Mobileye at this time. What is the reason?


  Combined with Intel's second-quarter financial report this year, revenue was US$15.321 billion, down 22% year-on-year; net loss was US$454 million, turning from profit to loss year-on-year, the first net loss in nearly 30 years. After the earnings report, Intel's stock price fell 10%.


  Also, Intel has struggled in the chip market. Not only will it be threatened by rivals such as AMD and Apple, but the total CPU market share has also dropped from nearly 80% a year ago to less than 65%.


  Gone are the days of "Intel inside".


  In the prospectus, Mobileye said the IPO proceeds will be used for working capital and general corporate purposes, and to repay some of Intel's debt.


  That is to say, the purpose of Intel's spin-off of Mobileye this time is not only for the prospect of autonomous driving, but also for self-help.


  However, it is not so easy to use subsidiaries to fight fires, and Mobileye's position in the market is also at risk.


  Forerunners lack strength


  Despite the constant rumors, it is undeniable that Mobileye is still the leader in the ADAS market.


  Since its establishment in 1999, Mobileye has quickly won most of the market share in the ADAS field, with a global market share of up to 70%, including Nissan, Cadillac , BMW, GM, Volkswagen and other auto manufacturers have deployed Mobileye's ADAS solutions. For a long time, the other players only got their share.


  Behind the high market share, on the one hand, for some car companies that do not have the ability to self-research algorithms, Mobileye is the best choice for "getting on" autonomous driving. In addition, Mobileye cooperates with Tier-1 manufacturers by providing self-developed chips and algorithms. After packaging the chips and algorithms, Mobileye provides autonomous driving solutions to OEMs, thereby accumulating a huge customer base.

In terms of financial data, Mobileye’s prospectus shows that in the first half of 2022, Mobileye’s revenue was $854 million, and in fiscal 2021, Mobileye’s revenue was $1.386 billion, a year-on-year increase of 43%. Revenues were $879 million and $967 million, respectively. From 2019 to 2021, Mobileye's EyeQ SoC shipments were 17.5 million, 19.7 million, and 28.1 million, respectively. The actual shipment in the first half of 2022 was 16 million, and shipments are also increasing year after year.


  In 2021, Mobileye will be at its peak.

But after the debut of the new car-making force, the situation of the Mobileye family's dominance began to change.


  Although Mobileye still maintains a leading position in the industry in terms of shipments and revenue, it has become an indisputable fact that it is not favored by the new forces of car manufacturers.


  In 2016, Tesla took the lead in abandoning Mobileye. Since then, "Wei Xiaoli" who adopted the Mobileye EyeQ4 chip also left it and turned into the arms of NVIDIA Orin. In addition to the new car-making forces, even traditional major customers such as BMW, Volkswagen, and GM have successively announced their "breakup" with Mobileye.


  Mobileye was "disgusted" because the previous "black box" ADAS solution was no longer viable for new energy smart vehicles.


  In order to achieve higher perception capabilities, most smart cars use multi-sensor fusion perception solutions, and the increase in the number of sensors also increases the requirements for computing power. The computing power requirement of L2 autonomous driving only requires 2-2.5TOPS, but the computing power requirement of L3 autonomous driving requires 20-30TOPS, and it needs more than 200TOPS to L4.

Comparing the chips of different companies, it can be found that Mobileye's EyeQ chip does not have an advantage in computing power. Taking EyeQ5 as an example, the computing power of a single chip is 24TOPS, while NVIDIA's Orin single computing power reached 254TOPS in the same period. Although the newly launched EyeQ Ultra has reached the computing power of 176TOPS, it will not be available until the end of 2023. At present, the multi-sensor fusion solutions of car companies such as NIO, Zhiji, and Xiaopeng have relatively high requirements on computing power. For example, Xiaopeng’s G9 computing power has reached 508Tops, and the computing power of NIO ES7 has also reached 1016TOPS. , if you use the Mobileye solution, you need to stack more chips.


  At the same time, the long iteration cycle is also what Mobileye is criticized for. It is reported that the iteration speed of Mobileye EyeQ chip is 3-4 years, and there is a gap of 4 years between the release time of EyeQ3 and EyeQ4. Nvidia's product update cycle is 2 years, and Qualcomm's is 1 year.


  With the deepening of automobile intelligence, users have also put forward a lot of personalized needs. Facing the needs of users, car companies cannot rely on the same set of solutions if they want to differentiate themselves. After 2020, with the new car-making forces such as Ideal and Xiaopeng abandoning Mobileye successively, the unbundling of software and hardware has also become a key word for car companies.


  And this is undoubtedly a fatal blow for Mobileye, which is highly coupled with software and hardware. Because Mobileye's closed black box solution does not support car companies to modify the algorithms, and vehicle data is not shared with car companies, before Mobileye and Jikr provided the highway driving assistance software package to Jikr 001 through OTA this year, many times Car companies cannot independently upgrade software through Mobileye's black box solution, which greatly limits the freedom of car companies.

According to SAIC Chairman Chen Hong's "Soul Theory", autonomous driving capability is the core competitiveness of car companies, and Mobileye's bundled sales model is tantamount to holding the lifeblood of car companies in their hands. , especially dangerous.


  Cheng also Xiao He defeated Xiao He. Mobileye started out because of the business model of software and hardware packaged and sold, and now it is deeply shackled by this model.


  Insufficient computing power, long iteration cycle, and high coupling of software and hardware have resulted in the lack of stamina for Mobileye, a pioneer. It is not difficult to understand why Mobileye's valuation will be adjusted again and again: the capital market does not buy it, not because its current situation is not good, but because it is not optimistic about its future expectations.


  Changes to Mobileye


  Times have changed, and Mobileye has to wake up from a sweet dream.


  Mobileye, which has lost customers, has fallen behind in technology, and is deeply stuck in development bottlenecks, is naturally aware of the current unfavorable situation and has tried to make changes.


  As early as 2018, Mobileye founder Amnon Shashua announced that starting from the next chip EyeQ5, Mobileye's business model will be changed from closed to open. Judging from the subsequent release of EyeQ5, this open version of the chip does support programmable operations.


  At the 2022 CES exhibition at the beginning of this year, Mobileye launched the EyeQ Ultra system integrated chip with a computing power of 176TOPS. It adopts a 5nm process technology and consumes less than 100W. The main goal of this chip is to achieve L4 autonomous driving.

 At the same time, there are two new EyeQ SoCs, EyeQ 6L and EyeQ 6H. The former is aimed at L1~L2 autonomous driving, while the latter focuses on solving all L2+ ADAS functions, and can also support third-party applications including visual parking and driver monitoring.


  It is worth noting that in order to allow car companies to gain more flexibility in the EyeQ 6H and EyeQ Ultra chips, in July this year, Mobileye released the EyeQ Kit, a software development kit, to provide car companies with a more open ecosystem. and rich development tools.

Specifically, EyeQ Kit enables car companies to develop software freely on the platform, and supports the embedding of third-party application software, thereby reducing costs. In addition, automakers can use the EyeQ Kit to gain access to a full suite of Mobileye solutions including computer vision, road information management (REM) autonomous vehicle mapping technology and other driving strategies.


  Simply put, this toolkit allows car companies to obtain complete services based on Mobileye’s years of technology accumulation, and can be customized and developed according to user needs to achieve differentiation.


  Based on these changes, it is not difficult to see that Mobileye is gradually letting go, and gradually transforming from a pure "chip + algorithm" supplier to a high-level autonomous driving travel service provider.


  But judging from the current situation, except that the new car jointly developed by Krypton and Mobileye will be equipped with six EyeQ 5 system integrated chips, there is no news about other car companies and Tier 1 equipped with EyeQ 6 and EyeQ Ultra chips. .


  It’s not good to be closed. Now that customers are open, they still don’t buy it. Where is the problem?


  First of all, Mobileye's openness is limited, and the role of EyeQ Kit is limited to the two newly released chips, EyeQ 6H and EyeQ Ultra, while EyeQ 6L is mainly for the L1-L2 assisted driving market.


  Second, the current business model of autonomous driving solutions has also changed. The mainstream solution is that car companies choose sensors and algorithms provided by different suppliers or self-developed algorithms after selecting chips based on product performance and cost considerations. For example, the ideal autonomous driving solution is NVIDIA Orin + Hesai LiDAR + self-developed autonomous driving algorithm. Although Mobileye has decoupled software and hardware, the problem is that many car companies have insufficient algorithm development capabilities, and even if Mobileye provides tools, they will not be able to use or not work well.

Image: Nvidia Orin

In addition, Mobileye's changes are slightly "long overdue". At present, NVIDIA's Orin has achieved mass production, and not long ago, its Thor chip with a computing power of up to 2000TOPS has also come out. Mobileye's EyeQ Ultra, which can meet L4 autonomous driving, will not be available until the end of 2023.


  It's too late to wake up, how should Mobileye find new breaking points? For Mobileye, the night ahead may be even longer.


  AB side of the coin


  The former king fell to the altar, and the plot of "Scarring Zhongyong" was repeated on Mobileye, but at the same time, through the rise and fall of Mobileye, we can also see that the pattern of the entire industry is being reshaped.


                                                    Pictured: Intel CEO Henry Kissinger

  From an international point of view, the previous "three-legged" situation of Mobileye, Nvidia and Qualcomm began to disintegrate, and Nvidia's momentum was particularly strong. Although NVIDIA can't beat Mobileye in terms of market share for the time being, from the perspective of computing power and customer acceptance, NVIDIA's performance on autonomous driving chips is even brighter. In addition to sitting firmly on the throne of chips in the smart cockpit, Qualcomm has also been actively deploying autonomous driving chips in the past two years. It is reported that all Volkswagen car brands will use Qualcomm self-driving SoCs from 2026.

 Mobileye's prospectus shows that China is Mobileye's second largest revenue region after the United States, but in recent years, the development of China's autonomous driving chip industry has not been inferior. At present, Huawei's MDC intelligent driving computing platform has released four products with different computing power, MDC210, MDC300, MDC610 and MDC810. BAIC Polar Fox HI, Great Wall Auto Salon Mecha Dragon, and Nezha S all use the MDC 810 computing platform, with a computing power of up to 400TOPS.


  In addition, the domestic core forces represented by Horizon, Black Sesame Intelligence and Xinchi Technology also have the momentum of "shooting the front waves on the beach". Take Horizon's newly released Journey 5 as an example. After the release of Journey 5, the Ideal L8 Pro became the first model in the world to adopt the Horizon Journey 5, and the L7 Pro will also be equipped with the Horizon Journey 5. In addition, the Journey 5 also won BYD and FAW Hongqi's models. Order.


  Old rivals are chasing after them, new players continue to exert their strength, and Mobileye is struggling.


  So why is Intel sending Mobileye to the stock market at this time?


  Backed by the big Intel tree, Mobileye's old capital can still eat for several years. But the fact is that Intel’s revenue in the second quarter of this year was US$15.321 billion, a year-on-year decrease of 22%. It was the first time in 30 years that it had a net loss. After the financial report was announced, Intel’s stock price fell by 10%. And in order to cut costs, Intel is planning to lay off thousands of jobs recently.


  On the contrary, with the good momentum of autonomous driving, Mobileye's revenue in the second quarter of 2022 increased by 41% year-on-year to $460 million, which is Intel's fastest-growing business.


  However, Mobileye's state is not optimistic. Mobileye posted a net loss of $75 million in the fiscal year ended December 25, 2021, compared with $328 million and $196 million in the last two fiscal years.


  There may be two reasons for Mobileye to be spun off and listed at this time: one is to understand Intel's urgent needs; the other is to use the power of capital to accelerate the development of Mobileye before the large-scale application of autonomous driving.


  "For Intel, this is a very right time. Now the market has very high growth expectations for EyeQ5, autonomous driving platforms and other businesses. Therefore, Intel can get the most from the market's high expectations." Analysis by Morgan Stanley Teacher Harlan Sur wrote in the research report.


  In fact, as early as December last year, Intel had already announced plans to spin off Mobileye. At that time, Intel CEO Henry Kissinger said that most of the funds raised will be used to build more chip factories. In addition to building factories, the independent Mobileye can also ease Intel's financial pressure. In this prospectus, Mobileye stated that the IPO proceeds will be used for working capital and general corporate purposes, and to repay some of Intel's debt.

From Intel's point of view, it has been losing ground in the "game" with AMD, and AMD is occupying more and more server and PC market share. In the competition with Nvidia, the latter also took away a lot of the cake in the GPU field, resulting in Intel's living space being gradually squeezed.

Therefore, for Intel, it is the top priority to spend money on R&D.


  However, although Mobileye is also difficult to protect itself, after all, the thin dead camel is bigger than the horse, and Mobileye itself also has many advantages and market opportunities.

At this stage, the focus of autonomous driving chips has changed from emphasizing large computing power to more cost-effectiveness. Mobileye, which started with a "low-cost, low-power" cost-effective solution, still maintains a fine tradition of doing its best on the premise of ensuring performance. Possibly reduce power consumption and cost. For example, the EyeQ 6L, the follow-up product of the EYEQ 4, is only 55% of the price of the EYEQ 4. Although the power consumption information has not been disclosed yet, it is officially said that this is an ultra-low power consumption chip.


  In addition, Mobileye is also actively deploying L4 autonomous driving. At the beginning of the year, it announced a comprehensive cooperation with Krypton, and plans to launch the world's first L4 autonomous vehicle in 2024.


Post a Comment