Tesla CEO Elon Musk has revealed the finalized design of the company’s next-generation AI5 chip for the first time, but in a surprising shift, the new hardware won’t be powering its future vehicles as we had expected.
Musk revealed the AI5 chip in an early morning post on X, congratulating Tesla’s AI chip team on the accomplishment, as well as preemptively thanking TSMC and Samsung for bringing the chip to production. Musk later added that he had been working on the chip “every weekend without exception for over 6 months.”
Congrats to the @Tesla_AI chip design team on taping out AI5!
— Elon Musk (@elonmusk) April 15, 2026
AI6, Dojo3 & other exciting chips in work. pic.twitter.com/hm54TdIzBx
In semiconductor terms, “taping out” means the design phase is complete and the chip is ready to move into production, a critical step that locks in the architecture and performance capabilities. But perhaps the most notable takeaway wasn’t the milestone itself, it was Musk’s clarification about its intended use.
Ever since AI5 was first introduced, it was widely expected to power Tesla’s next-generation vehicles. With recent development milestones, it was assumed it would debut in the upcoming Cybercab. However, based on the latest design and production timelines, AI5 isn’t expected to reach mass production until 2027. That meant the Cybercab, which is anticipated to begin production this month, would launch using the existing AI4 hardware.
Even last month Musk was talking about how AI5 is “optimized for Robotaxi.” And two months ago he was saying AI5 “will make the cars almost perfect.”
Now he has seemingly abandoned those plans. In response to a question asking whether these chips will be “go into cars or the robots,” Musk said AI5 will be instead only be deployed in Optimus and its growing fleet of AI training supercomputers.
Musk didn’t clarify whether AI5 might eventually end up in vehicles, only saying that Tesla’s current AI4 hardware is already sufficient for “much better than human safety for FSD.”
Optimus and our supercomputer clusters.
— Elon Musk (@elonmusk) April 15, 2026
AI4 is enough to achieve much better than human safety for FSD.
That response drew immediate skepticism from Tesla fans who questioned whether AI4 would be sufficient for FSD Unsupervised, given how Musk has previously claimed the same for HW3 (and older) vehicles, which are now unable to run the latest version of FSD (Supervised). Owners of HW3 vehicles have been stuck on FSD v12 for nearly a year and a half, although the company has previously said a “v14 lite” version would be released in Q2 2026.
Musk has already said the company would upgrade HW3 owners to newer self-driving computers due to its inability to run the latest models. However, those retrofits won’t happen until FSD Unsupervised is achieved.
There is of course the possibility that Elon Musk is saying this to avoid the “Osborne effect,” where customers delay purchases in anticipation of a newer, better product just around the corner. By saying that AI4 is already capable of delivering on Tesla’s autonomy goals, Musk may be trying to keep demand steady for its current lineup rather than risk buyers holding off for hardware that is still years away from production.
As we noted above, the AI5 chip is expected to enter early production in 2026, with full-scale manufacturing ramping up in 2027. Musk has also said the company is already working on AI6, AI7, and even AI8, with plans to dramatically shorten development cycles, targeting a chip design cadence of as little as nine months.
Always get the latest - make Drive Tesla your preferred source on Google
Are you buying a Tesla? If you enjoy our content and we helped in your decision, use our referral link to get three months of Full Self-Driving (FSD).Drive Tesla uses affiliate links, which means we may earn a commission if you make a purchase through them, at no additional cost to you. Thank you for your support.
