Particle.news
Download on the App Store

Musk Unveils Terafab in Austin With Goal of Producing One Terawatt of AI Compute

No schedule accompanied the space-first, all-in-one fab plan.

Overview

  • Tesla, SpaceX and xAI launched Terafab as a joint initiative in Austin, starting with an Advanced Technology Fab near Tesla’s Gigafactory, with Musk later saying it will comprise two fabs each dedicated to a single chip design.
  • The project targets two chip families: an edge/inference chip for Tesla vehicles and Optimus robots, and a hardened “D3” chip intended for AI data centers in orbit.
  • Musk set an ambition of one terawatt of compute per year and argued that current global AI compute is roughly 20 gigawatts, saying existing foundries can cover only a small share of his companies’ projected needs.
  • The facility is pitched as a fully integrated operation—design, lithography masks, fabrication, memory, advanced packaging and testing under one roof—to enable faster iteration.
  • Reports peg initial costs around $20–$25 billion with no firm construction or production timeline, and industry observers highlight multi‑year build times, steep capital demands and reliance on SpaceX’s Starship to cut launch costs toward $100–$200 per kilogram for space deployment.