Terafab Chip Plans: Musk Sets Test Date for Texas Fab

Terafab Chip Plans finally have dates attached. Elon Musk told analysts this week that the Tesla and SpaceX chipmaking venture is aiming to produce its first working test chip around the middle of next year at a research facility in Austin, with volume manufacturing targeted for the third quarter of 2028 at a far larger site in rural Texas, according to the Tesla News Wire's account of the call.

The project is Musk's attempt to build one of the most advanced semiconductor operations on the planet. The long-term goal, he has said, is to manufacture enough advanced AI chips each year to support about a terawatt of computing power — chips destined for Tesla's vehicles and humanoid robots, plus SpaceX's planned orbital computing systems. On social media this week he went further, claiming the value of the fab's eventual output would place a single building among the top twenty national economies in the world.

The timeline Musk laid out

According to the transcript reported by EVwire, Musk said the Austin research lab will begin installing machinery in the first half of next year, hoping to produce "just a kind of SRAM chip or something to basically test the line" around May or June. A simple memory chip is the logical first step: it is one of the easiest circuits to manufacture and verify, making it the standard way to shake out a new process before committing to the far more complex logic chips that would eventually power products like the Optimus robot.

Musk attached heavy caveats to both dates. He said he puts large error bars on the internal targets and called them extremely aggressive timeframes for chip production, acknowledging that his team is still learning fabrication from scratch. The company's finance chief, Vaibhav Taneja, added that meaningful output would more realistically arrive late in the decade, and that Tesla has already placed most of the equipment orders for the research fab.

The venture is structured as a joint effort between the two companies, with Intel participating as a technology partner. Tesla is building the first research fab near its existing Austin footprint, while SpaceX is developing the larger Grimes County complex at a site called Gibbons Creek, where land clearing was already visible by late September. According to an Everyday Next explainer, the long-term blueprint describes more than one hundred million square feet of manufacturing space — all of it still a goal, not a finished factory.

The bet reflects how central chips have become to Musk's empire. Tesla needs custom silicon for driver-assistance computers and the coming wave of Optimus units, while SpaceX wants radiation-tolerant processors for Starlink satellites and its planned space-based data centers. Buying that capacity from outside foundries means competing with every other AI lab for fab slots — building it yourself means competing with no one, if you can pull it off.

What it will cost — and who pays

The money involved is staggering. Asked how much capital the project needs, Musk said it would take tens of billions of dollars of equipment through the first months of that year and probably more than one hundred billion dollars over the long run. How the bill splits between Tesla and SpaceX has not been decided. County records reviewed by Engineering News-Record outline a potential multi-phase expansion reaching one hundred nineteen billion dollars, while the first phase alone represents nearly seventeen billion dollars of announced investment.

The schedule also depends on suppliers. Musk thanked the Dutch equipment giant ASML for speeding up deliveries, though he would not confirm whether the project has ordered the extreme-ultraviolet lithography machines that define cutting-edge chipmaking. Separately, the head of Taiwanese chipmaker TSMC confirmed this month that his company sent teams to evaluate a Texas expansion and discussed possible participation, while insisting any fab running TSMC technology would have to be operated by TSMC — a condition not yet reconciled with Musk's plan for his own companies to run the site.

For anyone who uses AI tools daily, the Terafab Chip Plans matter because chips are the bottleneck. Musk argues that controlling memory, packaging and logic in-house will let Tesla scale its humanoid robots and its autonomous Cybercab fleet without waiting on outside foundries. The same supply-chain logic sits behind the export controls reshaping who can buy advanced AI hardware — see our coverage of China's move to lock down top AI models. And the race to secure compute keeps intensifying: dormant AI agents are already straining the infrastructure they run on.

Skeptics note that building a leading-edge fab typically takes established chipmakers years of painstaking yield work, and Musk's manufacturing history includes several timelines that slipped badly. If the Terafab Chip Plans hold to schedule, though, next summer's test chip will be the first verifiable signal — either proof that a car company can learn chipmaking at record speed, or another reminder that atoms are harder to move than bits.