Automotive industry: a massive AI-powered transformation is ongoing

ADAS to AD sensor & computing market is showing a 12.1% CAGR between 2021 and 2027. Yole Développement (Yole) announces a US$25 billion market at the end of the period.

The market research and strategy consulting company Yole Développement (Yole) and its partner System Plus Consulting have been analyzing the semiconductor industry for several years, with a particular focus on automotive applications. In 2035, C.A.S.E will be a US$318 billion market, and the semiconductor value in a car will reach US$78.5 billion in 2026, with a 14.75% CAGR between 2020 and 2026, announces Yole in its Automotive Semiconductor Trends report… Those figures point out the crucial role of automotive players in the development of semiconductor innovations and business opportunities.

All year long, Yole and System Plus Consulting analysts deliver a comprehensive understanding of the industry, combining technical trends and market evolution. Based on a dedicated methodology and high-added value expertise in the semiconductor technologies applied in the automotive sector, the two companies publish numerous analyses, including technology and market reports, quarterly market monitors, and dedicated teardown tracks throughout the year. More information.
It took ten years to happen… ADAS technologies used to be considered a ‘nice to have’ feature for safety-conscious family car owners. It has now become the new frontier for technology companies like Intel, Qualcomm, Huawei, Apple, Google, and Tesla… So, what has happened and what is next to come?

Pierre Cambou, Yole Développement


Pierre Cambou, Principal Analyst Imaging at Yole, comments: “The C.A.S.E megatrends have taken the industry by storm, the whole ecosystem including automotive companies, semiconductor manufacturers… is moving to adapt to this massive transformation.”

The reorganization of the industry

Mobileye, from Intel, filling for an initial public offering at an estimated value of US$50 billion is a sign of the times when the same number was achieved by Xilinx, acquired by Intel’s competitor, AMD, in February 2022. A few months back, Waymo, the autonomous vehicle arm of Google, was raising US$2.3 billion at a value of US$30 billion while Huawei was shifting 10,000 employees from mobile activities, hit by US sanctions, to autonomous vehicle technology research. Other examples can be mentioned: BMW willing to fight off Tesla by partnering with Qualcomm and Arriver™, the perception venture born from the US$4.5 billion acquisition of Veoneer… More news available on i-Micronews.com.

There is an endless list of noteworthy news pointing toward industry reorganization. Therefore, all semiconductor and technology giants are looking into the ADAS to AD transition, the new center of gravity of the AI revolution.
Road mobility is a US$10 trillion market. ADAS to AD technology could be the magic ring to command it all.
This includes cameras, radars, LiDARs, and a lot of edge and cloud computing. Yole’s automotive analysts predict ADAS to AD technology to grow at 12% CAGR for the 2021-2027 period, reaching, at the electronic module level, US$25 billion in 2027.

“Today, only 12.3% of the 1,300 million cars on the road worldwide are fitted with ADAS technology,” explains Pierrick Boulay, Senior Technology & Market Analyst, Lighting and Display at Yole. “This figure will rise to 49% of the 1,800 million cars on the road estimated for 2030. This evolution is due to the huge increase in penetration of the technology in cars produced every year. From 21% five years ago, ADAS is now fitted onto 65% of all vehicle production in 2022, and this number will reach 86% by 2027.”

The long road to autonomy

While quantitatively the transition is almost total, the qualitatively new generations of hardware are dubbed with ‘Level of autonomy’ increments, originally Level 1.
Level 1 includes improved cruise control that could provide warnings to drivers. Then a transition came to Level 2 in different flavors: Level 2+ and now Level 2++ for cars with automated steering and braking, under supervision; in certain ideal conditions, and we cannot yet call it “autonomous driving”, but it is clearly at the top end of “advanced driving assistance”, states Yole’s analysts.
The excitement is therefore palpable at Level 2. Key enabling technologies that benefit from this evolution are cameras, radars, LiDARs, and computing chips.

In the Imaging for Automotive report, Yole’s imaging team describes the US$440 worth of electronic content needed for Level 2+ cars of 2017, which is jumping toward US$3,270 for state-of-the-art Level 2++ cars such as top-end Tesla, the recently released Honda Legend, and upcoming Mercedes S-class. The threshold toward completely removing the eyes of the driver from the road is nearby. This is the indication that Level 3 has been reached, the Kàrmàn line for autonomy which then extends to Level 4 and even Level 5. While autonomy will be a long road for cars sold to the public, electronic module expenditure for ADAS to AD technology will double in the next five years. The primary ADAS technology workhorse so far is radar, which is expected to reach US$9.3 billion, and then camera modules including an image sensor, optics, and a camera enclosure will represent US$8.9 billion. In parallel, LiDAR, the long-awaited technology breakthrough for autonomy, should represent US$1.7 billion, while the vision processor which powers all the smart ADAS features will represent US$5.1 billion in 2027.
Automotive perception has become a significant market, and advances in technology involve semiconductor-related hardware and software. The nexus of competition revolves around imaging, irrespective of the modality; radar startups are all striving for the new generation 4D imaging radar, while LiDAR startups have been promising 3D imaging Lidar under US$100 for many years now.
In the meantime, imaging using standard CIS is being pushed to the limits of its perception capabilities. ‘Brute force’ are the words used by some companies to describe the high-computation-power strategy developed by AI companies to solve AD challenges through ‘camera only’ approaches.
Other players diversify the modalities. Indeed, the list is growing notably through the introduction of Lidars and thermal imaging solutions, also possibly SWIR imaging solutions in the near future, as detailed in the SWIR Imaging report from Yole.

Adrien Sanchez - Yole Développement

“But one thing is certain,” comments Adrien Sanchez, Technology & Market Analyst, Computing at Yole. “AD technology is not totally there yet for automotive consumers, while robotic taxi fleets using a different set of hardware and software are starting to expand in a selected number of cities. At a figure of US$100K per car, there are already a few thousand robotic cars doing the job. AI technology combined with extremely powerful chipsets will also require different sensor modalities; all these are the main factors favoring autonomy.”

‘More Moore’ and ‘More than Moore’ are indeed racing in a formidable ‘Chicken Run’.

Legacy automotive players vs. consumer players. Who will win?

On the road to driverless and electric cars, disruptive and new players are entering the automotive industry. For several years, semiconductor companies like Mobileye, Nvidia, and Qualcomm have positioned themselves at the center of automated driving systems. These companies supply the chips necessary to process data from cameras, radars, and LiDARs.
At the same time, players from the consumer market, such as Huawei, Alibaba, Baidu, and Sony, among others, are entering the automotive industry. Most of these players are partnering with established OEMs to build a new car and supply the technology for autonomous / automated driving. The recent partnership between Honda and Sony clearly illustrates this trend. Honda has the knowledge and the resources to manufacture the car, and Sony will bring its expertise in the development and application of imaging, sensing, telecommunication, network, and entertainment technologies.
In the last two years, many new OEMs appeared in the automotive landscape. They are building electric cars around semiconductors and AI software for self-driving features. A closer look at their sensor setup reveals a distinct disruption compared to legacy OEMs. While luxury cars from Mercedes or Audi remain quite conservative in their camera setup, new players, like XPeng, Nio, Link&Co, IM motors, and Arcfox, are integrating an average of six ADAS cameras plus four viewing cameras per car.
And that is only for the exterior cameras. To this must be added two in-cabin cameras, five radars, and up to five LiDARs…
The transformation of the automotive industry is massive. This is a fact. But no one really knows who will win and what will happen at the end of the road trip. Nevertheless, all players are in pursuit of the C.A.S.E megatrends in which sensors and AI computing are center stage. Solving the “imaging for automotive” equation will probably make the difference. How many cameras, radars, LiDARs, and so on, what processing power, and at what cost is perhaps all that matters…

Yole and System Plus Consulting analysts are deeply engaged in this new paradigm. Make sure to follow our teams to get a valuable understanding of this evolution and strategic moves of the ecosystems.

Related reports

Imaging for Automotive 2022
Towards 20 cameras in upcoming cars, imaging for automotive will reach $25B in 2027.



Automotive Semiconductor Trends 2021

Automotive Semiconductor Trends 2021
For the first time in its history, the automotive industry must face new industrial and technological challenges while undergoing dramatic changes in its value chain.

Ai for Automotive 2020

Artificial Intelligence Computing for Automotive 2020
Artificial Intelligence paves the way to full autonomy.



Source: http://www.yole.fr/, https://www.systemplus.fr/

Related presentations

Cet article vous a plu ?

Partagez-le sur vos réseaux sociaux