Musk bets Tesla's AI future on Intel node that isn't finished yet
EV maker leaning on still-in-development 14A process for Terafab, says it needs to build own silicon
Elon Musk used Tesla's latest earnings call to reveal plans to build AI chips on Intel's not-yet-finished 14A process – a bet on silicon that doesn't exist.
"Intel is excited to partner with us on some of the core manufacturing technologies," Musk said, referring to Tesla's planned "Terafab" chipmaking push, which he says is aimed at producing its own AI silicon at scale. He added that Tesla plans to use "Intel's 14A process, which is state-of-the-art and in fact, not yet totally complete."
Elon Musk wants to build 50 times more chips than the world currently produces, using 'new physics'
READ MOREThat last bit didn't seem to trouble him. "By the time Terafab scales up, 14A will be probably fairly mature or ready for prime time… we think it's going to be a great partnership," Musk added.
The 14A process Musk is referring to is a future node beyond Intel's upcoming 18A, itself not yet in production, that's supposed to be the chipmaker's shot at clawing back relevance against rivals that have spent the past few years eating its lunch.
Musk framed the move less as opportunism and more as existential panic about supply. "Terafab is not some sort of mechanism to generate leverage over our chip suppliers," he said. "We don't see a path to having enough sufficient quantity of AI chips down the road… we just anticipate hitting the wall if we don't make chips ourselves."
That urgency doesn't quite line up with the numbers [PDF]. Tesla is still shifting a huge number of cars, but revenue dipped year-on-year, and profits fell further, with margins getting squeezed as costs rise and spending on AI and manufacturing keeps climbing.
Which helps explain the rest of the pitch. If the next phase of Tesla is built on autonomy and AI rather than just selling cars, then securing its own supply of chips – or building them outright – starts to make sense.
- Intel gets trapped in Elon's reality distortion field as it joins in megafab delusions
- Musk makes the Macrohard joke again
- Musk distracts from struggling car biz with fantastical promise to make 1 million humanoid robots a year
- Tesla revenue falls for first time as Musk bets big on robots and autonomy
There's also the usual Muskian moonshot layered on top. "We do have some ideas for how to make maybe radically better AI chips," he said, quickly caveating that these are "research ideas… long shot," albeit ones that could deliver a "giant improvement" if they land.
That same long-shot thinking runs through the rest of Tesla's plans. On robotaxis, he stuck to the familiar script: expansion is coming, slowly but surely, with Europe still a tougher regulatory nut to crack. On humanoid robots, the pitch was even bigger. Musk again positioned Optimus as the endgame, repeating that it could become Tesla's "biggest product," with production ramping – in his words – "slowly" before eventually scaling to something meaningful.
Put it together and Tesla looks less like a car company branching into AI, and more like something else entirely that still happens to sell cars. Tesla is preparing to spend tens of billions to build chips, robots, and autonomous systems, while warning it may not have enough silicon to keep up with its own plans.
And if that sounds like a lot to pull off at once, Musk appears unfazed – after all, the manufacturing process he's betting on isn't even finished yet. ®
