The planned facility is designed to consolidate multiple stages of chip production into a single site, including logic and memory fabrication, packaging, testing, and lithography mask production. Musk said this configuration would enable faster iteration cycles by eliminating the need to move wafers between separate facilities during development and testing.
TeraFab is expected to produce two categories of chips. One will focus on edge inference workloads for Tesla’s vehicles and its Optimus humanoid robots. The second will be engineered for space environments, where higher operating temperatures can reduce the need for additional cooling hardware on satellites.
Musk tied the project to long-term compute demand across his companies, including Tesla, SpaceX, and xAI, which SpaceX acquired earlier this year. He estimated current global AI compute output at roughly 20 gigawatts per year, describing that level as only a small fraction of projected needs. He suggested terrestrial demand could reach 100 to 200 gigawatts annually, with significantly higher capacity required for space-based systems powered by solar energy.
Despite the investment, Musk said his companies will continue sourcing chips from existing manufacturers such as TSMC, Samsung, and Micron, while encouraging those suppliers to expand production.
The announcement did not include a timeline for when the Austin facility would begin operations or reach full output. Musk has previously referenced advanced process nodes for the project, but did not specify a target in the latest remarks.
By bringing chip design, manufacturing, and testing under one roof, the project reflects an effort to reduce reliance on external suppliers as demand for AI and compute infrastructure accelerates.
This analysis is based on reporting from Tom's Hardware.
Image courtesy of Tesla/SpaceX
This article was generated with AI assistance and reviewed for accuracy and quality.
