CES 2026: Power Electronics Innovations Fueling AI and Electric Vehicles

The Lynk & Co 06 Relive compact SUV will use an A-PHY-spec SerDes chipset. (Image: Lynk X Co.)

Physical AI may have been the dominant keyword at CES 2026, but behind all the hype around that, there were still plenty of companies on hand in Las Vegas focused on less flashy headlines. Here are some examples of companies working on the power electronics required behind the scenes to make our potential artificial intelligence future possible.

Valens

Valens announced during CES that a global premium automaker that sells vehicles in China would be the next to use its VA7000 MIPI A-PHY-compliant chipsets. Valens said this was the fourth design win for an A-PHY chipset, which “reinforces the connectivity standard as a frontrunner for next-generation ADAS and autonomous systems.” The vehicle that uses the VA7000 chipsets will go into production in 2027.

Valens called MIPI A-PHY the first standardized solution for high-speed sensor connectivity, and noted that it is the only one with design wins across multiple silicon suppliers. In September 2025, the MIPI Alliance announced that Geely Auto Group would use an A-PHY-spec SerDes chipset in mass production in the Lynk & Co 06 Relive compact SUV.

Valens showcased its evolving MIPI A-PHY ecosystem during CES, including multiple A-PHY-enabled products. The company is working on medical imaging and machine vision devices that can use the same chipset from the automotive industry with the same capabilities.

Tim Wendel, Valens’ Director of Product Marketing, told SAE Media that the reason the A-PHY chipset is gaining traction inside and outside the auto industry is “the ultimate resilience of our technology.”

Valens calls the MIPI A-PHY chipset the frontrunner for next-generation ADAS and autonomous systems and that it is the first standardized solution for high-speed sensor connectivity with design wins across multiple silicon suppliers. (Image: Lynk X Co.)

“If you have a higher data rate, the more susceptible you are against electromagnetic noise, which can cause failures in the car, especially when you talk about autonomous or partly autonomous driving vehicles,” he said. “You want data to arrive safe, so that if a compute system wants to analyze video feeds, you don’t want this to be corrupted or interrupted.”

Valens’ transmission cables and strategy are also well-prepared for future EVs with zonal architectures, Wendel said. As the distance between sensors and compute grows, longer cables serve as antennae that pick up and send noise, among other issues.

“As sensors become higher resolution — so you talk about five, 8, 12 megapixels — that means data rate increases,” Wendel said. “You disturb others more. You need heavy shielding in order to protect your link, but also to not disturb others with your irradiated emission. At the same time, you have mobile phones and 5G inside the vehicle and radar and more outside the vehicle, it’s a unique collection of noise that dynamically changes while you’re driving, and all that is a huge risk to the car.”

Silanna

Like Valens, Silanna is attempting to broaden the market for its products outside of automotive. Silanna recently introduced a laser driver design called FirePower that integrates power, firing, and fault-sensing functions on a single chip. FirePower can drive sub-2ns laser pulses with a peak power of up to 1000W and a 10 MHz pulse repetition frequency. Silanna said its LiDAR sensors could be used in hunting sights, cycling safety cameras, and golfing computers, as well as in self-driving vehicles.

Mark Drucker, Silanna President and CEO, told SAE Media at CES that Silanna’s RF business was acquired by Qualcomm in 2019, after which the company focused more on power management, including AC-to-DC and DC-to-DC converters, and then research into GaN technologies. Silanna got into the LiDAR market based on this background and its experience with lasers. With all of this knowledge, Drucker said Silanna wants to help Tier Ones and OEMs lower the cost of LiDAR, not just be a commodity player.

“We can’t win that way,” Drucker said. “We want to be a value-added supplier. In the LiDAR space, cost is the real barrier that’s been preventing it from being widely adopted. Slowly but surely, it’s starting to get there. Eliminating a lot of the mechanical scanning, MEMS technologies, all of that, and getting to a truly solid state scan LiDAR system is one of the steps the industry wants to be able to take to get there, but they haven’t fully resolved all those challenges.”

One of Silanna’s answers to the challenge of bringing down the cost of LiDAR sensors is work on laser array stacks. Rajeev Thakur, Silanna’s Director of Marketing and Business Development, told SAE Media that its time-of-flight sensor customers are looking for anywhere from 24 to 56 lasers on one die.

“This will obviously help to bring the cost down, and you’re still able to individually fire and control them,” he said. “However, when you go to a stack of lasers like that, firing them is not easy, because when you fire one, there tends to be some leakage. Our IP is basically that we can fire these arrays of lasers at high peak power and very low pulse width.”

Thakur said Silanna is working on drivers and GaN FET arrays that will allow these arrays to have the necessary current to fire the lasers.

Another answer Silanna is working on is improved analog-to-digital processing for frequency-modulated continuous wave (FMCW) LiDAR.

“As you know, FMCW is coming,” Thakur said. “That’s one of the weaknesses, I would say, for the Chinese OEMs, is they don’t have FMCW. They may be working on it. I’m pretty sure they are working on it, but they don’t have it out yet. So that’s an Achilles heel, if you will, for them. For FMCW LiDAR, you have a continuous wave, and when the wave comes back, you take that analog wave and make it a digital wave, then you can do the processing for that. And to have a very good wave with high resolution, you need a very high sampling rate. And this is where Silanna comes in. We can provide high-samplingrate analog-to-digital converters.”

Omni Design Technologies

Omni Design Technologies also believes that the automotive industry is moving toward FMCW LiDAR architecture. At CES, the company displayed some of its next-generation wideband signal processing (WSP) technologies for ADAS and wireless communications, as well as data center networking and space-based cellular networks. Like Silanna, Omni Design supports both FMCW and time-of-flight technologies.

Omni’s WSP multi-channel, synchronized analog-to-digital converters can simultaneously acquire and digitize data from multiple sensors, including radar, LiDAR, cameras, and ultrasonic for a better understanding of the world around the car. Omni Design said its Swift ADC can offer better object detection in less-than-ideal lighting and environmental conditions, thereby improving autonomous emergency braking and lane keep assist ADAS functions.

AIStorm

AIStorm says its mission is to redefine semiconductors through charge-domain processing. AIStorm has said that its charge-domain compute technology outperforms static ram-based process-in-memory because it uses fewer computational cycles and has lower power requirements. During CES, AIStorm said that, compared to a transistor, its charge domain process has a potential power improvement of up to 117 times and a footprint that’s smaller by up to 30 times. The end result would be more AI computations while using less silicon for faster, cooler, and smaller devices.

This article was written by Sebastian Blanco, Editor-in-Chief, Automotive Engineering magazine, SAE Media Group.