Why self-driving car company Cruise is bringing chip design in-house

Last week, Cruise announced a new DIY project: in-house chip design.

The General Motors-owned self-driving startup is following in Tesla’s footsteps by no longer relying fully on Nvidia for chips—and beginning to develop its own processors that will be deployed in Cruise vehicles by 2025.

"Two years ago, we were paying a lot of money for a GPU from a famous vendor," Carl Jenkins, head of hardware at Cruise, told Reuters. He added, "There is no negotiation because we're tiny volume. We couldn't negotiate at all. So that's why I said, ‘Okay, then we have to take control of our own destiny.’”

But when it comes to chip development, taking control of one’s own destiny is expensive. The type of AI chips likely needed to power Cruise’s autonomous vehicles typically cost over $100 million to design, according to Jared Mondschein, a physical scientist at the RAND Corporation.

Cruise expects the chips to help it bring down vehicle costs, and it plans to scale up production to help pay for the strategic move, Reuters reported. The strategy aligns with GM’s ultimate plan to “democratize hands-free driving across vehicle segments and price points,” as Darryll Harrison Jr., GM’s director of global EV/AV, told us last year.

Cruise’s in-house silicon team has developed four chips to date, including a “main brain” processor for its Origin vehicle, which focuses on data processing and using machine learning to reduce latency, and another chip built for sensor data, like radar, lidar, camera input, and acoustics.

The company is partnering with a “leading manufacturer in Asia to produce them at scale,” Ann Gui, Cruise’s head of silicon said in a statement from Cruise spokesperson Rachel Holm. Holm declined to name the partner or specify how much Cruise has invested in in-house chip development so far.

Under pressure

Cruise’s decision to move chip design in-house is part of a larger shift among automakers and tech companies more broadly—tech giants like Apple, Google, Microsoft, and Meta have invested in in-house chip design as well. By 2025, Gartner predicts, half of the top 10 automotive original equipment manufacturers will move the practice in-house.

And with chip design software like electronic design automation (EDA) growing more advanced, there’s a lower barrier to entry for companies looking to make the switch.

“I’d be surprised if they weren’t relying on EDA tools—just because designing these computer chips is extremely expensive and time-consuming, and it’s so complex that you need these EDA tools to really be able to do it efficiently,” Mondschein said. Holm confirmed that Cruise does use EDA tools.

In 2018, Tesla became the first automaker to make the switch.

“When there weren't sufficient numbers of…chips available, Tesla was able to redesign their chips to match the supply chains that were available, and then they did some software reengineering to align their software systems with the new chips,” Mondschein said. “They were able to avoid the supply-chain crunch that the other manufacturers experienced. So I think a lot of other companies have taken note of that and realized that moving the process in-house allows for greater flexibility to react to supply-chain constraints.”

Those constraints are becoming the “new normal,” Murat Aksel, chief procurement officer at VW, told German media outlet Automobilwoche.

“Overall, the situation has improved, but only in small steps," he said. “Investments for new capacity are now on track, but there will probably still be a structural shortfall in semiconductors up to and including 2023.”

Mondschein predicted that following Tesla’s success and Cruise’s announcement, more automakers will switch to in-house development in coming years—that is, if they have the upfront capital to invest in design costs.

“A lot of it will depend on how supply-chain constraints shake out, if the companies making the switch now are successful,” Mondschein said. “Cost is a major factor here, and especially the type of chip these…companies are trying to produce will also matter.”

Update: This piece has been updated to clarify that Cruise will continue to work with Nvidia in some other capacities.