Musk targets 1 terawatt of AI compute from new Texas chip plant by 2028
Tesla and SpaceX plan a massive semiconductor facility in Gibbons Creek that aims to produce enough processing power to rival the GDP of a mid-sized nation.
By Robert Hayes ·
Elon Musk announced on October 20 that critical semiconductor suppliers, including ASML, are accelerating equipment deliveries for the Terafab manufacturing project. This joint venture between Tesla and SpaceX is designed to produce high-volume chips locally. The announcement came during a live broadcast of SpaceX's third-quarter 2026 earnings call on X. Musk stated the goal is to reach volume production at the primary site in Gibbons Creek, Texas, by the third or fourth quarter of 2028. He described this schedule as highly aspirational and noted it carries significant uncertainty.
The path to this timeline involves several key milestones. In April 2026, Intel joined the venture to co-develop silicon fabrication technology. This partnership provides access to the 14A manufacturing process, which Terafab will use with traditional extreme ultraviolet lithography systems. ASML, the sole commercial supplier of these $400 million EUV machines, is considered the most critical partner in the supply chain. Unofficial sources previously suggested the first EUV tool might arrive in mid-2027, though it is unclear if this aligns with the expedited timeline Musk referenced.
Other major equipment providers include American firms Applied Materials, Lam Research, and KLA Corporation, along with Japan's Tokyo Electron. These companies will supply front-end processing and metrology machinery. Tesla and SpaceX are also negotiating with approximately 10 South Korean manufacturers, including Hanmi Semiconductor and HPSP. These firms provide advanced packaging equipment and high-pressure hydrogen annealing tools. The project aims to vertically integrate logic, memory, and advanced packaging in a single facility.
The scale of the operation is intended to address future shortfalls in AI processors from external foundries like TSMC. The chips produced at Terafab are earmarked for xAI clusters, self-driving vehicles, Optimus humanoid robots, and SpaceX's orbital AI data center. Musk claimed that once fully operational, the commercial value of the chip output from this single building could place it within the top 20 global country GDPs. For comparison, Switzerland, the 20th largest economy, has a GDP of roughly $1.14 trillion.
Musk outlined broader computing ambitions during the same broadcast. He expects his AI companies to operate 2.5 to 3 gigawatts of capacity by the end of 2026. He expressed high confidence in adding another 10 gigawatts in 2027, aiming for a total of 13 gigawatts of AI compute by year-end. To achieve this, SpaceX and xAI are aggressively procuring next-generation NVIDIA architecture. Plans include bringing a 440,000-GPU Vera Rubin training cluster online by the second quarter of 2027.
The 13 gigawatt target contrasts with other major projects. The restructured 10 gigawatt Project Stargate and Meta's 5 gigawatt Hyperion are both expected to reach full capacity in the 2030s. SpaceX also plans to manufacture its first experimental test chip at its research fabrication facility in Austin, Texas, around the second quarter of 2027. This serves as a precursor to the full-scale production at Gibbons Creek.
Skepticism remains regarding the 1 terawatt annual output target for Terafab. Gartner estimates the total global AI-optimized data center capacity is roughly 44 gigawatts as of late 2026. This figure represents only 4.4 percent of Terafab's planned output. While global AI computing capacity has grown significantly since 2022, experts question whether the supply chain can support such an aggressive expansion from a single site.
Musk emphasized that the company is appreciative of suppliers accelerating deliveries to meet these aggressive dates. The project represents a multi-billion dollar investment in a vertically integrated chip factory. It seeks to internalize the production of advanced architectures that are currently dependent on specialized global suppliers. The success of the timeline will depend heavily on the coordinated efforts of these diverse partners.
Source: Tom's Hardware
