Musk’s Terafab: A $1 Trillion Gamble on Texas Chip Manufacturing
Tesla and SpaceX are pushing suppliers to accelerate equipment deliveries for a massive new factory aimed at producing one terawatt of AI compute annually by the end of the decade.
By Gabriela Flores ·
The semiconductor industry is facing a potential paradigm shift as Tesla and SpaceX announce that key equipment suppliers have agreed to fast-track deliveries for their joint Terafab project. This move signals a significant escalation in the race to secure domestic production capacity for artificial intelligence processors.
During a live broadcast of SpaceX’s third-quarter earnings call on October 20, Elon Musk confirmed that ASML and other critical partners are expediting shipments. This acceleration is designed to support an aggressive timeline for the facility, which is located in Gibbons Creek, Texas. The primary goal is to reach volume chip production by the third or fourth quarter of 2028, a deadline Musk described as highly aspirational with significant margin for error.
The project’s technical backbone relies on a collaboration with Intel, which joined the venture in April 2026. The facility will utilize Intel’s 14A manufacturing node technology alongside traditional extreme ultraviolet lithography systems from ASML. ASML is the sole commercial provider of the $400 million EUV machines required for leading-edge production, making their cooperation essential for the project's success. While one EUV tool was previously slated for mid-2027 delivery, it remains unclear if the recent expedited schedule refers to this specific unit or additional tools.
Beyond lithography, the supply chain involves major American equipment manufacturers such as Applied Materials, Lam Research, and KLA Corporation, as well as Japan’s Tokyo Electron. These firms will provide front-end processing and metrology machinery. Additionally, Tesla and SpaceX are in negotiations with approximately ten South Korean companies, including Hanmi Semiconductor and HPSP, to source advanced packaging and high-pressure hydrogen annealing equipment.
The commercial scale of Terafab is intended to rival that of major global economies. Musk stated that once fully operational, the facility’s output could place it within the top 20 global GDPs, comparable to Switzerland’s economy of roughly $1.14 trillion. The factory aims to fabricate logic, memory, and advanced packaging internally, targeting an annual output of one terawatt of AI computing power. This capacity is designed to supply chips for xAI clusters, Tesla’s autonomous vehicles, Optimus robots, and SpaceX’s orbital data centers.
In parallel with the manufacturing push, Musk outlined near-term computing goals for his AI entities. He expects 2.5 to 3 gigawatts of active capacity by the end of 2026 and aims for a total of 13 gigawatts by the end of 2027. To achieve this, SpaceX and xAI are procuring Nvidia’s next-generation "Vera Rubin" architecture, with plans to bring a 440,000-GPU training cluster online by the second quarter of 2027. This is notably faster than the projected timelines for competing projects like Project Stargate and Meta’s Hyperion, which are expected to reach full capacity in the 2030s.
Despite the ambitious scope, industry experts have raised questions about the feasibility of the one-terawatt target. Gartner estimates that global AI-optimized data center capacity is currently around 44 gigawatts, a figure that represents only 4.4% of Terafab’s planned annual output. While global AI capacity has grown by roughly 3.3 to 3.4 times annually since 2022 according to EpochAI, the gap between current infrastructure and Terafab’s goals remains substantial. The project ultimately seeks to mitigate reliance on external foundries like TSMC, reducing future shortages of cutting-edge processors.
Source: Tom's Hardware
