Musk unveils Terafab — a Tesla and SpaceX chip factory targeting 1T of space-based compute
What's the deal? Elon MuskDealroom has a profile for this one. Try Dealroom → has unveiled Terafab, a joint chip manufacturing venture between Tesla, SpaceX, and xAIDealroom has a profile for this one. Try Dealroom →, to be built near Tesla's Austin headquarters. Speaking at a former power plant in Austin, Musk said the project is a necessity: "We either build the Terafab or we don't have the chips, and we need the chips, so we build the Terafab." He set a target of producing chips supporting 100 to 200 gigawatts of computing capacity annually on Earth, and eventually a terawatt in space. No timeline was given.
Musk envisions two chip types: one for Tesla's vehicles and Optimus humanoid robots, and one called D3 designed for space environments. The space chip would power solar-driven AI satellites in low orbit, which Musk argues will eventually be cheaper to run than terrestrial data centres.
Why now? Both Tesla and SpaceX are voracious consumers of AI compute — for autonomous driving, robot training, satellite processing, and avionics — and the global shortage of high-performance chips has tightened supply and pushed delivery timelines out. Tesla has previous form here: it developed its own D1 training chip for its Dojo supercomputer. Musk framed Terafab as the logical next step in that vertical integration strategy, extended to cover the full chip stack.
What could go wrong? Building a leading-edge semiconductor fab is one of the most capital-intensive industrial undertakings in existence. Recent US foundry projects have cost tens of billions of dollars with multi-year timelines, and require specialised talent — process engineers, lithography experts — that is already in short supply globally. Musk has no background in semiconductor manufacturing and a well-documented history of setting aggressive timelines that slip.
The more plausible near-term path, analysts suggest, is a hybrid approach: sourcing advanced logic wafers from established foundries while building domestic capacity in packaging, test, and specialised process nodes for automotive and space-grade parts. Whether Musk's announcement reflects that measured ambition or something more grandiose remains unclear.
The signal: Terafab reflects a broader push by the world's largest AI consumers to secure their own compute supply chains rather than depend on a handful of foundries. Nvidia, TSMC, and SamsungDealroom has a profile for this one. Try Dealroom → sit at a bottleneck that every major AI player is now trying to route around — through acquisitions, partnerships, or, in Musk's case, the prospect of building from scratch. Whether Terafab becomes operational infrastructure or remains a vision statement will depend on concrete milestones: site filings, equipment orders, and hiring. Until those materialise, the gap between Musk's ambition and silicon on shelves remains very wide.
Sources:
Bloomberg
Business Insider
TechCrunch
A.M.
