Initial prototype fab operations are to be focused in Austin, Texas, in proximity to Tesla's existing Gigafactory Texas. In May 2026, SpaceX estimated an initial investment of US$55 billion and a total investment of US$119 billion for all phases of the prototype fab.[3][2][1][4] The full-scale Terafab is to be built at a yet to be determined location. Analysts estimate the costs for the full-scale facility at between US$5−13 trillion.[5][6]
Background and development
AI chips for Tesla Autopilot are among the products planned for manufacture at Terafab.
The project was first teased by Musk in early 2026 and officially announced on 21 March 2026 during a special event at the defunct Seaholm Power Plant in Austin, Texas.[4] Musk described Terafab as a pivotal step toward humanity becoming a galactic civilization, with the announcement accompanied by a live SpaceX broadcast on X and conceptual imagery of prototype 100 kW "AI Mini Sat."[2] He went on to describe that the global chip industry cannot expand quickly enough to meet the demand that Tesla will need for "edge inference compute" for Tesla vehicle and Optimus humanoid robot production, nor for the special semiconductor characteristics required for orbital AI infrastructure. Musk claimed all the current fabrication facilities on Earth produce only about 2% of what Tesla and SpaceX will need across all projects,[4] saying "We either build the Terafab, or we don't have the chips, and we need the chips, so we build the Terafab."[7] The project integrates efforts under the SpaceX/xAI umbrella with Tesla's existing silicon development.
On 7 April 2026, Intel announced it would join the Terafab project to contribute manufacturing expertise.[6] As of 2026, Intel is one of three manufacturers worldwide who are producing sub-5 nm chips at scale, the other two being TSMC and Samsung Electronics. In April 2026, Musk announced Intel's 14A manufacturing process would be used at the full-scale Terafab, claiming by the time Terafab scales up, the process "will be probably fairly mature or ready for prime time".[5] He also announced a division of labour, with Tesla focusing on the prototype fab and SpaceX in the lead on the initial part of the full-scale Terafab.[5]
Musk said the long term goal is to have 1 million wafer starts per month and produce between 100 and 200 billion custom AI and memory chips per year in the full scale facility.[4]
Prototype fabrication facility
Tesla plans to build a prototype "Advanced Technology Fabrication" facility at the existing Tesla GigaTexas site that will be capable of producing each of the parts of chip manufacturing in one facility in order to iterate rapidly—"make a chip, test it, revise the mask, and repeat without shipping wafers between sites"—a capability that does not currently exist in any other chip fab site globally. The aim is to manufacture chips for both AI edge inference and for AI model training with chips optimized for operation in space.[7]
The project targets 2-nanometer process technology and an initial output of 100,000 wafer starts per month,[4] though this was later revised by Musk to "maybe a few thousand wafers per month, but it's really intended to try out ideas".[5] Tesla's fifth-generation AI chip, AI5, is among the first products the pilot facility will be designed to produce, with small-batch production anticipated in 2026 and volume production in 2027.[8]
.jpg)
