One of the most interesting questions in the auto-tech sector is how long Mobileye will continue to dominate it. The Israeli company, acquired by Intel for $15.3 billion in 2017, is a source of pride, but constitutes an anomaly from a broader business and technological perspective. Through tier-1 suppliers, who connect the auto manufacturers to the smaller suppliers, Mobileye directly or indirectly holds over 70% of the global market in advanced driver-assistance systems (ADAS) - one of the richest and fastest growing markets in the auto industry.
Mobileye has led the market for five years, in contrast to any known scenario in high tech in general and the auto sector in particular, in which at least one strategic competitive threat arises during this time span. It seems, however, that it is not easy to overcome Mobileye's many years of experience in machine vision and video processing algorithms, and certainly not when the company itself is still reinventing itself and enjoys access to Intel's resources.
The job of figuring out who will threaten Mobileye is even harder and more frustrating, because the auto industry does a good job of keeping its secrets, for reasons of competition. Furthermore, the Israeli auto-tech sector is mostly managed by ex-IDF officers from the Intelligence Corps and other classified units of the defense forces. These ex-officers take the obsession for field security implanted in them during their military service into civilian life; they are stingy in disclosing information.
History shows that market leaders cannot forever ignore their competitors' power (ask Microsoft about Google). In this case, the elephant in the room is Toyota, the world's largest auto manufacturer, which sells over 11 million cars a year. Toyota's budget and technology resources are inexhaustible, and it resolutely refuses to adopt Mobileye's technology. Toyota has always claimed that it could achieve better results independently than those of Mobileye, and therefore has no need to tie itself to the Israeli company's closed system. Since Toyota has close ties with a number of Japanese tier-1 suppliers, it can mobilize the awesome development resources of the entire Japanese industry for its needs.
Toyota's conservative engineering and business practices have also slowed its progress in the autonomous vehicle segment. In recent weeks however, the company has pulled a very daring timetable out of its sleeve. At last month's CES exhibition, Toyota displayed for the first time an autonomous prototype based on a Lexus model. This week, the company announced that it planned to commercially market an autonomous car next year under the Lexus brand. This car will probably have Level 4 autonomy - independent driving, but on specific routes and in specific conditions, not a car that needs neither a steering wheel nor a driver.
The announcement nevertheless signals a big step forward. Senior executives in Toyota's division for advanced technologies R&D this week referred to the company's future vehicle as a "super computer on wheels," with processing power that will leave all of its competitors far behind. There have been many grandiose pronouncements in this matter in the past two years, including from Elon Musk, who claimed that he had the world's strongest chip for an autonomous vehicle.
It turns out that Toyota's impressive breakthrough comes from the Israeli auto-tech market: a company named Cortica.
Overcoming the machine learning barriers
Cortica appeared in an in-depth "Globes" story two years ago. A veteran company by Israeli high-tech standards, Cortica has worked in video processing and images utilizing artificial intelligence (AI) with government customers and intelligence agencies in Israel and throughout the world. The company was founded on the basis of theoretical research about the human brain at Technion, Israel Institute of Technology. Its approach upsets the prevailing order in AI.
Cortica asserts that in its accepted form, deep learning, the dominant AI technology, is a kind of black box lacking transparency, although such transparency is essential for analyzing the reason for an accident involving an autonomous vehicle, for example. Cortica says that this approach wastes computer resources and is not efficient enough for an "environmental concept" and fusing sensors in real time - currently the main technological obstacle in the way of an autonomous vehicle.
Cortica says that its technology takes the accepted form of deep learning a step further. It facilitates unsupervised self-learning methodology with "flat hierarchical architecture at low power" that makes it fully transparent in processing and self-learning and unguided capability. This is no simple matter, considering that ordinary deep learning has been supported by hundreds of billions of dollars from venture capital investors and huge companies like Intel. The fact that Cortica has thus far refused to disclose its auto industry partner, and that its technology is considered enigmatic by technology experts was certainly not helpful.
A few weeks ago, however, the company announced that it had installed its algorithm on a super-chip for an autonomous vehicle of Japanese tier-1 auto supplier Renesas as a solution for Level 2 and 3 ADAS autonomous vehicle systems, which are extremely common in today's market. At the same time, the two companies are already working on a more advanced generation of the chip.
This is where things become more interesting. Renesas, whose name is unknown outside the industry, is a huge company known as the world's largest supplier of microcontrollers. Its vehicle division offers a comprehensive array of solutions for the global auto industry. Only recently, Renesas acquired and merged into itself a US competitor in a $6.7 billion deal.
Since 2017, the company has set its sights on becoming the global leader in ADAS systems based on machine vision. In February 2018, it made its first move in this direction by launching a "super computer on a chip" for machine vision processing, called R-CAR V3, designed to compete directly with Mobileye-Intel's EYEQ4, for high-level autonomous vehicles.
Mass production of the chip will begin in the third quarter of 2019. It will allegedly feature "a tiny fraction of the electrical consumption of the competing chips." The announcement was greeted with skepticism by the auto industry, because the main problem of a "super computer for a vehicle" of this type is an insatiable appetite for electric power, something that is particularly challenging for electric cars. Claims of a "jump in performance, combined with a dramatic reduction in power consumption" are usually self-contradictory.
Renesas did not say in 2018 how this breakthrough was achieved, but a joint announcement by Cortica and Renesas in January 2019 made several striking claims: "By integrating Cortica's autonomous AI technology into the system of Renesas's chip system, we have overcome the existing technological obstacles of safety for every watt of electricity in an autonomous vehicle. Beyond this, our concept solves the known barriers of ordinary deep learning." The companies also said that Cortica's technology, which is implanted on Renesas's chip, "requires only a tenth of the processing power required for Mobileye's system. This declaration directly challenges the dominant force in the ADAS market. It could be regarded as pretentious, an invitation to trouble, and/or self-confidence backed by orders from strategic customers.
Will Cortica play a role in Toyota's autonomous car?
In October 2017, Toyota and tier-1 suppliers Dentsu and Renesas announced a tripartite alliance for developing advanced systems for an autonomous vehicle. The announcement said that Renesas would supply its new R-CAR super chip to Toyota - the same chip on which Cortica's AI technology was installed. Starting in 2020, this chip will be the brain of the company's family of autonomous vehicles, together with Dentsu's control systems.
If the jigsaw puzzle pieces are assembled, the interesting conclusion is that Toyota's autonomous vehicle will be based on Cortica's extraordinary AI technology. This is certainly likely to one of the most important breakthroughs of the Israeli auto-tech sectors since Mobileye. Taking into account that the alliance is backed by Japanese companies with an aggregate market cap of $200 billion, it is doubtful whether Mobileye and Intel will be able to treat this threat with the same nonchalance they have been showing towards their competitors.
In any case, the importance of this development transcends the auto industry. If Cortica's "upgraded" AI technology passes the proof of feasibility stage of manufacturers and multiplies the accepted ratio of computer power per watt, it is likely to completely disrupt the AI industry, which currently amounts to tens of billions of dollars. In this case, we will not be surprised in Cortica becomes an object for very aggressive merger and acquisition attempts.
Cortica declined to respond.
Published by Globes, Israel business news - en.globes.co.il - on February 13, 2019
© Copyright of Globes Publisher Itonut (1983) Ltd. 2019