CES: chipmakers accelerate the cars of the future

Posted on Jan 5, 2023, 4:37 PM

At the Consumer Electronics Show (CES), which is being held this week in Las Vegas, automakers and chipmakers set up their booths nearby. Sign of the times. The former, like Nvidia, Qualcomm, Mobileye (Intel), need the latter to catch up with Tesla, the pioneer of smart cars controlled by a central computer. In contrast, the latter see the automobile as a huge driver of growth.

ARM, for example, announced that its turnover in the automotive sector will double from 2020. For its part, Nvidia recorded an 80% increase in its turnover in the last quarter in autonomous vehicles, and forecasts 11 billion dollars in revenue over the next six years from this segment.

According to S&P Global, the automotive semiconductor market will grow from $67 billion (2022) to $129 billion (2028). This market – which includes processors for autonomous and connected vehicles but also chips dedicated to optimizing electric motor batteries – has more than doubled since 2006.

At the opening of CES on Tuesday, Nvidia announced a partnership with Taiwanese electronics giant Foxconn. The latter is setting up a platform for creating the “computer brain” of its manufacturer customers’ connected cars. He will also start building his own cars. The iPhone assembler may also, in the uncertain future, produce autonomous cars that Apple is alleged to have designed in the greatest secrecy.

Tesla’s centralized model

This Nvidia Foxconn collaboration is emblematic of the microprocessor giant’s investment in the automotive market. Nvidia, which has been making chips for cars for twenty years, will offer Foxconn all its expertise in autonomous driving. The Taiwanese will thus gain strength in the sale of centralized car scanning solutions, which become “smartphones on wheels”. “We offer a central computer for the vehicle. All functions become software-based. For example, the front camera will see that it is raining, and the computer will decide to activate the wipers,” says Danny Shapiro, car manager at Nvidia. The challenge, he continues, is to “reduce the number of chips and therefore simplify the production chain”.

The Nvidia Drive platform, already used by manufacturers such as Mercedes-Benz, is open, which means they can write their own software on it. This model of the central computer is the one that Tesla developed for ten years – Nvidia was also a supplier to the pioneer of electric cars until Elon Musk decided to internalize the production of chips.

“Tesla has a large fleet, which collects data and allows it to add improvements. Other manufacturers take their cue from them. This is a good thing, because soon consumers will realize that it makes no sense to buy cars that cannot be updated, because they are not driven by software,” commented Danny Shapiro.

Reduce costs and increase performance

Also well positioned in the automotive sector, the American Qualcomm announced on Wednesday in Las Vegas that the new generation of microprocessors with an etching fineness of 4 nanometers will provide smart cars from 2025. The manufacturer also presented semiconductor its Snapdragon Ride Flex SoC processor, which combines on a single chip the management functions of “digital cockpit”, assisted driving, and autonomous driving.

For its part, Japan’s Sony has teamed up with compatriot Honda to launch a new electric vehicle brand that will use its expertise in image sensors and screens. The duo is currently only showing prototypes, but the first Afeela models are expected for 2026 on American roads.

For Wayne Lamm, consultant at CSS Insight, pioneers like Qualcomm are combining all the smart functions of the car on the same platform “not to minimize, because there is more space in the car than the smartphone, but to reduce costs and increase performance. .

More computing power

Shortages of semiconductors during the Covid crisis really penalized manufacturers who needed different types of chips for each vehicle. Instead of having one system for heating, one for windscreen wipers, one for locking the doors, it is a question of building a single brain, fed by an infinity of data that coming from sensors, radar, lidar, etc.

“As a result, conventional processors are no longer enough. Artificial intelligence pushes forward. We need to embed more computing power, more transistors, more advanced functions,” explains Wayne Lamm . The fun of a cross-country skier.

Solveig Godeluck (With LM and Fl.D.) (In Las Vegas)

Leave a Reply

Your email address will not be published. Required fields are marked *