Autonomous cars: the irruption of AI chip giants in a growing market

Artificial intelligence is disrupting the automotive industry. And the reciprocal is in the process of being verified. The observation made by Yole Développement may not come as a surprise. But it is difficult not to be impressed by the scale and speed of this upheaval exhibited in the latest study of the consultant specialist in semiconductors, carried out in partnership with System Plus Consulting and published on March 23.


“By 2027, all cars produced will be equipped with a driver assistance system (ADAS), compared to 65% of cars in 2021 and 21% five years earlier”, begins Pierre Cambou, principal analyst of the study, called “Automotive industry: a massive AI-powered transformation is ongoing”. The consequences for the car fleet are commensurate with this progression: if only 12.3% of the 1,300 million cars in circulation today incorporate an ADAS, this figure should reach 49% by 2030, on a fleet total estimated at 1,800 million cars.

2021-2027: annual growth of 12.1% for ADAS technologies

In addition, new vehicles placed on the market in the future should gradually increase their level of autonomy, marking the transformation from ADAS systems (levels 1, 2 and 2+) to semi-autonomous driving systems (with from level 3). “If we take the category of driving aids, for example, we have gone from simple lane keeping aid, to which has been added sign reading, and, more recently, emergency braking (AEB) with pedestrian detection. Today, we are developing autopilot functions on motorways. Each of these technologies requires increasingly complex sensors and ever more computing power,” details Pierrick Boulay, analyst specializing in light sensors at Yole.


Yole predicts that the ADAS systems integrated into new vehicles will continue to improve until 2050. @ Yole Development

By doing so, Yole estimates that the market for sensors and chips dedicated to ADAS and autonomous driving is expected to grow at a rate of 12.1% per year between 2021 and 2027 (in compound annual growth rate, or CAGR). “If we follow the S-curve of adoption of new technologies, well known from market analyses, this growth should slow down somewhat after 2027, but when we see the race for ever more innovative technologies that players in the ADAS, it is not certain that this slowdown will occur as the adoption of new autonomy (AD) technologies will take over”, adds Pierre Cambou.

With this evolution, the three technologies that are doing well in the personal car sector are cameras, radars and computing chips, note Yole analysts.


1649124255 916 Autonomous cars the irruption of AI chip giants in a


Yole predicts that the radar market will grow, between 2021 and 2027, from 5.6 to 15.3 billion dollars and that of cameras, from 4.6 to 8.9 billion dollars. @ Yole Development

The camera and the radar, the two legs of ADAS growth


“If you listen to Mobileye CEO Amnon Shashua or Tesla boss Elon Musk, they swear by the camera, says Pierre Cambou. For our part, we believe that the ADAS systems of the future will rely heavily on the camera-radar couple. »


1649124255 150 Autonomous cars the irruption of AI chip giants in a


The growth of the market for cameras intended for ADAS systems comes in particular from the fact that vehicles should integrate more and more of them as the systems improve, according to Yole’s estimates. @ Yole Development

Another category of sensors, lidars, will be increasingly used in a very special case, robot-taxis, which Pierre Cambou strongly distinguishes from the ADAS market, since “these are much more expensive vehicles”, but much fewer. “In ADAS, lidars will certainly be used, but marginally, to ensure redundancy with other sensors, and these will mainly be cheap lidars”, projects the analyst.

If all these sensors need ever more computing power as they accumulate and improve, it is above all the use of artificial intelligence algorithms, used to make vehicles more autonomous, which should boost the growing demand for chips, according to Yole.

Consequently, if the manufacturers of cameras (the American ON Semiconductor, the Chinese Omnivision, the French STMicroelectronics, the South Korean Samsung or even the Japanese Sony) and radars (dominated by Europeans such as NXP and STMIcroelectronics) should keep their leadership, “there are a lot of movements in the chip market for the self-driving car”, assures Pierre Cambou.

A power struggle in the AI ​​chip market for the autonomous vehicle

In this area, Intel’s bet to buy the Israeli Mobileye for 15 billion dollars in 2017, after a Tesla accident calmed the enthusiasm of the sector in 2015, “turned out to be a winner today”, indicates the analyst. Indeed, the company today represents 60% of the market, recorded a turnover of 1.4 million dollars in 2021 and, in preparation for an IPO in 2022, is valued at 50 billion. of dollars.

But it should have a lot to do, since the American giant AMD has just bought Xilinx, also present on the market, that Toshiba and Texas Instruments remain in the race and, above all, that a few newcomers are counting “to impose itself, like the American Ambarella or several Chinese, like Horizon Robotics”, continues Pierre Cambou.


A “power struggle » that the American chip giants for AI, such as Nvidia or Qualcomm, “ideally positioned” according to Yole’s analyst, are also joining. Qualcomm has, for example, just acquired Veonir, a Swedish tier 1 supplier, with whom it has launched a joint venture in the field, called Arriver. “Here, we are talking about the chips of the giants of the field, manufactured by leading companies in a semiconductor market which generated 550 billion dollars in 2021, against 5 billion for that of ADAS”, recalls Pierre Cambou. An interest that makes the expert say that the autonomous car will be “the next center of gravity of the AI ​​revolution”.

We would love to thank the writer of this short article for this remarkable web content

Autonomous cars: the irruption of AI chip giants in a growing market


You can view our social media profiles here as well as other pages related to them here.https://www.ai-magazine.com/related-pages/