Copyright Wccftech
In a move that would surprise a few in the semiconductor world, Tesla CEO Elon Musk announced plans for a "gigantic chip fab" to build Tesla's own AI chips, all while publicly mulling a new foundry partnership with Intel. Speaking at Tesla's annual shareholder meeting Thursday, Musk, fresh off securing a controversial $1 trillion pay package, laid out an aggressive new vision for Tesla's silicon. He stated that even with existing partners TSMC and Samsung, the supply of AI chips is "still not enough" to meet his ambitions for Full Self-Driving (FSD) and the Optimus humanoid robot. "I think we may have to do a Tesla terafab... it's like giga but way bigger," Musk said. "I can't see any other way to get to the volume of chips that we're looking for." This "terafab," according to Musk, would be a colossal facility with at least 100,000 wafer starts per month. But the most immediate bombshell was his open courting of Intel, a move that sent Intel's stock jumping 4% in after-hours trading. "You know, maybe we'll, we'll do something with Intel," Musk uttered to a cheering crowd. "We haven't signed any deal, but it's probably worth having discussions with Intel." For Intel Foundry Services (IFS), which has struggled to land a high-volume, next-generation anchor client, a potential Tesla deal would mean a monumental victory. If it happens, it would validate their high-stakes, multi-billion-dollar bet on becoming a world-class foundry and would go on to prove that major tech players are now seriously considering them as a primary alternative to TSMC and Samsung. Taking Direct Aim at NVIDIA's Dominance Musk's entire strategy seemingly is to present a clear alternate to NVIDIA's AI dominance. While he praised NVIDIA's hardware, he made it clear that Tesla "cannot be dependent on a single supplier" for the "tens of millions" of AI chips he needs. He then made a typically bold claim, estimating Tesla's custom-designed chip would be "inexpensive, power-efficient" and would "probably consume about a third of the power used by Nvidia's flagship Blackwell chip, at 10% of the cost to make." While these figures are highly ambitious, the message is clear: Musk is done with with being at the mercy of NVIDIA's high-margin, supply-constrained GPU market and is willing to spend billions to build a vertically integrated alternative. "I'm super hardcore on chips right now as you may be able to tell," Musk added. "I have chips on the brain." A New Chapter in the Foundry Wars Musk's announcement certainly doesn't mean he's dropping his existing partners. Instead, he appears to be building a multi-foundry "all of the above" strategy to ensure Tesla's supply chain is invulnerable. This new plan complements recent news, including: AI4 Chip: Currently in production at Samsung. AI5 Chip: Design is complete and will be dual-sourced from TSMC and Samsung. AI6 Chip: A massive $16.5 billion deal was recently signed with Samsung for production in its Taylor, Texas fab, with volume production expected in mid-2028. By bringing Intel into the fold as a potential partner and planning to build his own "terafab," Musk is all set to pit the world's top three foundries against each other. This creates a hyper-competitive environment that can drive down prices and guarantee Tesla gets the capacity it needs for its FSD and robotics ambitions.
