Elon Musk said late Wednesday that Tesla and SpaceX AI expect to keep buying Nvidia chips in large amounts. The comment drew fresh attention to the computing power Elon needs as he pushes deeper into self-driving cars, humanoid robots, and AI systems.
It also stood out because it was the first time Elon publicly used the name SpaceX AI after SpaceX bought xAI in an all-stock deal last month.
But the chip demand is tied to Tesla’s next in-house processor, called AI5. Tesla is building that chip for its self-driving plans. The processor is meant to run the company’s autonomous systems, including Full Self-Driving software.
In a separate post on X, Elon said AI5 can be used for training inside data centers, but said it is mainly built for edge AI work inside Optimus and Robotaxi. He also said Tesla expects a wide release of a Full Self-Driving (Supervised) update within a few weeks.
Last Saturday, he added that Tesla’s Terafab project to make AI chips would launch in seven days.
The new comments line up with what Elon said last year about how hard it is to get enough chips. At Tesla’s annual meeting, he said the company might work with Intel as it looks for more manufacturing options.
Elon said, “You know, maybe we’ll, we’ll do something with Intel. We haven’t signed any deal, but it’s probably worth having discussions with Intel.”
No deal has been announced, but the message was clear. Tesla is still looking at every path that could help it lock in more supply.
At the same time, Elon said Tesla was also working with TSMC in Taiwan and Samsung in South Korea. Even with those suppliers, he said the company still might not get enough volume.
At that annual meeting, Elon said, “Even when we extrapolate the best-case scenario for chip production from our suppliers, it’s still not enough.” He then laid out the next step in plain terms. He said:-
“So I think we may have to do a Tesla terafab. It’s like giga but way bigger. I can’t see any other way to get to the volume of chips that we’re looking for. So I think we’re probably going to have to build a gigantic chip fab. It’s got to be done.”
The AI5 chip is being built for cars, but Elon also tied it to Optimus and Robotaxi.
While Elon talked about buying more chips, Nvidia used its annual GTC conference to roll out new products and new auto partnerships. During his March 16 keynote, Jensen Huang announced updates to the company’s networking lineup.
Nvidia launched the Rubin platform, which includes six new chips built to power an AI supercomputer. The company also introduced Inference Context Memory Storage and more efficient Spectrum-X Ethernet Photonics switches, along with other products.
Meanwhile, Jensen also said that Nvidia is teaming up with Tesla’s rivals, Chinese carmakers BYD and Geely, as it tries to grow beyond core AI demand.
Speaking Monday at the GPU Technology Conference in San Jose, California, he said, “The ChatGPT moment of self-driving cars has arrived.”
Both Chinese automakers will use Nvidia Drive Hyperion, a platform that combines chips, computers, sensors, and software for Level 4 autonomous vehicles. These vehicles are designed to run without human control in specific areas or under certain conditions.
Isuzu and Nissan were also added to Nvidia’s robotaxi platform. Hyperion covers the full chain, from cloud-based AI model training to real-time decisions on the road.
The push comes as China’s EV market cools after a very strong December. Sales had jumped late in the year as buyers rushed before incentives ended. NEVs are no longer exempt from purchase tax this year, and the Chinese New Year in February added more pressure.
EV sales fell 32% year over year. The overall market was also weak, down 25% to around 1 million units. BEVs dropped 35% to 278,000 units. PHEVs fell 31%. EREVs held up better, down 16%, while non-plug models were down 19%. Large SUVs helped EREVs because that segment was hit less hard after the incentives ended.
Get seen where it counts. Advertise in Cryptopolitan Research and reach crypto’s sharpest investors and builders.