ADAS and autonomous vehicle, two different faces of the same trend

Autonomous vehicles, currently known as Robotaxis, are already in full operational mode in multiple locations in China and USA, providing a service considered pure science fiction 10 years ago. For sure, this is a limited experiment, and consolidation is happening due to extreme economic issues, but don’t underestimate the fundamental change brought by some players with a trend that is here to stay, a global change of what a car is for in our day-to-day lives, with more robotics entering our homes, companies, hospitals, and shops…

It is quite different if you are looking at the evolution of ADAS. Full autonomy for the car you can buy at a nearby shop is less and less an objective for car manufacturers. One of the main reasons is that such a revolution is taking way more time than initially expected, and the CAPEX and R&D budget needs are exploding. That is why car manufacturers are now focusing on the hottest issue, the transition to electric vehicles. So, it has become a matter of managing budget priority. But the need for medium levels of autonomy (2,2+,2++, even 3) is there, and even if full autonomy is no longer a clear objective, providing more assistance to the driver is clearly on the roadmap of all the car makers. Specific functions focus on highway driving, parking, traffic jam, etc., targeting specific aspects of the driver experience step by step, bringing real value in strictly-defined areas instead of big promises with unclear added value in order to help and bring more safety on the road.

This is clearly a different face of the same trend: one focuses on a complete change of paradigm, changing everything we know about human transportation, while the other focuses, step by step, on onboarding the consumer for a long journey with more valuable comfort and security. The entire industry is trending toward robotics, which necessitates an increasing amount of semiconductors, growing from $510 in 2021 to $912 per car in 2027, pushing the ADAS market to $65.3B in 2027 (according to Yole’s latest report, “Semiconductor trends in Automotive”) and a market for autonomous vehicles that is just starting.

This ADAS market is not a single monolith; it’s a constellation of sensors and computing chips. It is constituted of radar, visible and night vision cameras, LiDAR, ultrasonic sensors, and, of course, the related data processing units, either decentralized or increasingly centralized. According to Yole’s latest reports on these various products, radar will be a $9.3B market in 2027, LiDAR a $1.7B opportunity, and CIS an ever-expanding market reaching $9B.
But in addition to such sensing modules, data processing is also becoming key for ADAS (ignoring autonomous vehicles for the moment, for which computing is the most complex part of this revolution). Now, car makers need computing platforms that are much more performant than just an MCU. So they are using Intel/Mobileye-based computing power, Nvidia-adapted GPUs, FPGAs, … to process the data and enable systems to respond immediately to incidents. The architecture is also evolving, moving from decentralized systems to centralized (or at least more centralized) processing modules, able to provide shared resources and more capabilities with redundant architecture to maximize security.

All these trends are analyzed and described in the different reports Yole Group has published over the last months, and you will find there all the information needed for benefitting from these multiple evolutions and revolutions.