Elon Musk Says Tesla Model S & X Are Almost Sold Out—Here’s Why Demand Is Exploding

Austin, Texas—Elon Musk has pulled Intel into “Terafab,” a project that aims to build a first-of-its-kind semiconductor plant capable of churning out a full terawatt of AI compute every year—about 70 percent of what TSMC makes worldwide today. Intel will supply its bleeding-edge 18A process node, packaging know-how and foundry muscle for the $20-$25 billion pilot line, joining existing partners SpaceX, Tesla and xAI. Terafab’s blueprint collapses every stage of chipmaking—design, lithography, fabrication, memory and advanced packaging—into two massive facilities on the Giga Texas campus. The near-term goal is 100,000 wafer starts per month on 2-nm technology, scaling to one million wafers and up to 200 billion custom AI and memory chips annually by the end of the decade. Early products will include Tesla’s next-gen AI5 processor for autonomous vehicles and humanoid robots, plus high-performance accelerators destined for both ground data centers and a planned constellation of “AI Sat Mini” satellites run by SpaceX. Musk argues the build-out is unavoidable: “We either build the Terafab, or we don’t have the chips, and we need the chips.” The billionaire says his companies will require more than 2 terawatts of compute within a few years—orders of magnitude above current U.S. output. Intel, meanwhile, gains a flagship customer for its nascent foundry business as it battles TSMC for advanced logic share. Key questions remain. Intel has disclosed no dollar commitments or equity terms, and analysts warn execution risk is enormous: a single advanced fab typically costs $20 billion and takes years to qualify. Musk’s own hardware timelines—Starship launches, Optimus robot production, the steering-wheel-free Cybercab slated to start manufacturing next year—have slipped before. Still, the announcement underscores a broader reality: the scramble for AI silicon is becoming national-infrastructure-scale. With hyperscalers building their own power plants and chip demand outstripping supply, Terafab—whether or not it hits the terawatt mark—signals that Musk intends to control the entire AI supply chain from raw wafers to orbiting inference clusters, and that Intel is betting its foundry future on helping him do it.