Musk promises 1 terawatt of AI chips from single Texas factory

The most striking figure in Elon Musk’s recent announcement is the target of one terawatt (TW) of annual AI computing power to be produced by a single physical facility. This ambitious output goal for the Terafab project, a joint venture between Tesla and SpaceX, would make the site’s commercial value comparable to the GDP of a top-20 global economy. For context, Switzerland, currently ranked 20th in global GDP, stands at approximately $1.14 trillion. Musk described this timeline as "highly aspirational" with significant margin for error, but the scale of the operation marks a significant shift in how the industry approaches vertical integration.

The path to this milestone began with the official inclusion of Intel in the venture in April 2026. Intel’s role is to co-develop and refactor silicon fabrication technology, specifically providing access to its advanced 14A manufacturing node. This partnership is critical because Terafab plans to utilize traditional extreme ultraviolet (EUV) lithography systems. The project’s massive planned output has necessitated a broad coalition of partners across the semiconductor supply chain, moving beyond a single foundry model to an internally integrated production system that handles logic, memory, and advanced packaging all in one location.

According to statements made during a live broadcast of SpaceX’s Q3 2026 earnings call on October 20, suppliers are moving faster than previously scheduled. Musk noted that the company is "very appreciative" of ASML and other key vendors for accelerating equipment deliveries. ASML, the world’s sole commercial supplier of the $400 million EUV lithography machines required for leading-edge chip production, is considered the most critical partner. While unofficial sources previously suggested the first EUV tool was slated for mid-2027, it remains unclear if that specific date has been pulled forward due to the expedited efforts Musk referenced.

Other major equipment providers, including American firms Applied Materials, Lam Research, and KLA Corporation, as well as Japan’s Tokyo Electron, are contributing front-end processing, metrology, and process control machinery. Additionally, Tesla and SpaceX are in negotiations with approximately ten South Korean equipment manufacturers. These include Hanmi Semiconductor, known for advanced packaging, and HPSP, which specializes in high-pressure hydrogen annealing, alongside firms providing specialized piping and process valves. This diverse supplier base is essential for mitigating future shortfalls of cutting-edge AI processors that have historically relied on external foundries like TSMC.

The facility, located in Gibbons Creek, Texas, aims to achieve volume chip production by the third or fourth quarter of 2028. A preliminary step in this process involves SpaceX manufacturing its first experimental test chip at a research fabrication facility in Austin, Texas, around the second quarter of 2027. These chips are intended to power a wide array of Musk’s ventures, including xAI clusters, self-driving vehicles, Optimus humanoid robots, and SpaceX’s orbital AI data center.

Beyond the physical chip manufacturing, Musk outlined aggressive computing capacity goals for his AI companies. He expects 2.5 to 3 gigawatts (GW) of capacity to be active by the end of 2026 and stated he is "highly confident" about adding another 10 GW in 2027. This would bring the total to 13 GW by the end of 2027, a milestone that contrasts with other major projects like the 10 GW Project Stargate or Meta’s 5 GW Hyperion, both of which are not expected to reach full capacity until the 2030s. To support this, SpaceX and xAI are procuring next-generation NVIDIA architecture, with plans to bring a 440,000-GPU "Vera Rubin" training cluster online by the end of Q2 2027.

Despite the detailed planning and supplier commitments, industry experts have raised questions about the feasibility of the 1 TW ambition. Research firm Gartner estimates that total global AI-optimized data center capacity is roughly 44 GW as of late 2026. This means Terafab’s planned output would represent a 4.4% increase to the entire existing global capacity. While global AI computing capacity has grown by roughly 3.3 to 3.4 times per year since 2022, according to EpochAI, the sheer scale of Terafab’s target remains a significant challenge for the current supply chain.