Chip Equipment Makers Push Shipments for Musk's Factory

Elon Musk says he wants to build 1 terawatt of AI computing power a year from a single plant in Texas. That figure, part of his plan for Terafab, would place the site's output among the top 20 national economies worldwide. Switzerland currently sits at number 20 with a GDP around $1.14 trillion.

Musk shared the plans during SpaceX's Q3 2026 earnings call, carried on X on October 20, 2026. He said ASML and other equipment suppliers are pushing deliveries faster to meet the schedule. He also put a date on volume chip production at the main site in Gibbons Creek, Texas: the third or fourth quarter of 2028. He called those dates 'very aggressive' and said the timeline is highly aspirational with big error bars. He added, 'We're very appreciative of ASML and all of our other suppliers in accelerating equipment deliveries in order to achieve this outcome.'

Terafab is a joint effort from Tesla and SpaceX. The plan calls for a multi-billion-dollar facility that makes logic, memory and advanced packaging in one place. Intel joined the project in April 2026 to help with silicon fabrication technology and provide access to its 14A process. Terafab would use extreme ultraviolet lithography with Intel's 14A node. ASML is the only commercial supplier of the $400 million EUV tools used for high-volume, leading-edge production. Sources have said the first EUV tool for Terafab was due around mid-2027, though it is not clear if that is the timing Musk meant. Other suppliers named include Applied Materials, Lam Research, KLA Corporation and Tokyo Electron. Talks are also underway with about 10 South Korean equipment makers.

The first step in the chip work is set for SpaceX's research fabrication facility in Austin, Texas, where the company plans to build its first experimental test chip around Q2 2027. The plant in Gibbons Creek is meant to supply chips for xAI clusters, self-driving vehicles, Optimus humanoid robots, SpaceX's orbital AI data center, and the wider AI industry. The goal is to reduce reliance on outside foundries such as TSMC for cutting-edge AI processors.

On the AI compute side, Musk said his companies expect 2.5 to 3 gigawatts of computing capacity active by the end of 2026. He said he is confident about adding 10 gigawatts in 2027 and would be surprised if the total reaches 13 gigawatts by the end of 2027. SpaceX and xAI are buying NVIDIA 'Vera Rubin' systems, with plans for a 440,000-GPU training cluster online by the end of Q2 2027. Project Stargate, restructured to 10 gigawatts, is set to reach full capacity in the 2030s, and Meta's 5 gigawatt Hyperion has a similar timeline.

Questions remain around the size of the Terafab goal. Gartner estimates global AI-optimized data center capacity at roughly 44 gigawatts as of late 2026, or about 4.4% of the 1 terawatt Terafab would produce. Since 2022, global AI computing capacity has grown about 3.3x to 3.4x a year, according to EpochAI.

Next up, the project moves toward its first test chip in Austin around Q2 2027, while the main plant in Gibbons Creek aims for volume production in the second half of 2028. Suppliers are being asked to speed up deliveries to hold that schedule.