Elon Musk has now announced Terafab, a semiconductor manufacturing project jointly undertaken by Tesla, SpaceX, and xAI, at the decommissioned Seaholm Power Plant in downtown Austin. The facility, planned for the North Campus of Gigafactory Texas in Travis County, aims to produce chips at an unprecedented scale. Musk described it as âthe most epic chip building exercise in history by far,â with a target of eventually delivering up to one terawatt of annual computing powerâfar exceeding current global semiconductor output from any single site.
SpaceXAI + Tesla TERAFAB Project
Goal is a trillion watts of compute/year
Most must necessarily go to space, as US electricity is only 0.5TW https://t.co/hMtg9vNLcw
— Elon Musk (@elonmusk) March 22, 2026
As part of the project, Terafab will consolidate the full semiconductor supply chain under one roof, encompassing chip design, lithography, fabrication, memory production, advanced packaging, and testing. The project targets 2-nanometer process technologyâthe most advanced commercially available node, currently limited to TSMCâand plans to initially produce 100,000 wafers per month, scaling toward 1 million wafers per month. Musk outlined two primary chip categories – inference processors for terrestrial applications powering Tesla vehicles, Cybercab robotaxis, and Optimus humanoid robots, plus high-power, radiation-hardened variants for orbital AI satellites. He allocated roughly 80% of output to space-based compute and 20% to Earth, arguing that orbital solar power â five times more intense than on the surface â combined with thermal advantages, could make space-based inference cheaper than terrestrial alternatives within three years.
The announcement follows Musk’s repeated warnings about chip shortages constraining his companies’ ambitions. Tesla’s Full Self-Driving and Optimus programs, xAI’s Grok models, and SpaceX’s satellite constellation plans require massive compute resources that existing foundriesâprimarily TSMC, Samsung, and othersâcannot supply at the necessary pace or volume. It was said that global production meets only a fraction of projected needs, leaving his ecosystem vulnerable to supply constraints.
Terafab builds on prior efforts – Tesla has developed custom silicon for vehicles and robots, while SpaceX pursues orbital data centers (including an FCC filing earlier this year for up to one million satellites). xAI, now a SpaceX subsidiary following its acquisition last month, will consume the majority of output for training and inference. Musk framed the project as essential to a future of âamazing abundance,â with billions of robots, interplanetary travel, lunar cities, Mars colonization, and galactic expansionâall dependent on vastly more compute than today’s industry can deliver.
From what we know so far, the facility will begin as an âadvanced technology fabâ in Austin, focusing on rapid iteration and testing before full-scale expansion. Musk projected annual production of 100â200 billion custom AI and memory chips for Earth applications, with additional capacity for space. No firm timeline was provided for full operations, though Musk suggested initial contributions to next-generation A15 inference chips (volume production targeted for next year) and the ongoing development of A16 chips. Still, the road ahead won’t be easy, since building at 2nm requires extreme precision, specialized equipment, and expertise. If successful, Terafab could vertically integrate Musk’s ecosystem, reducing dependence on external foundries and accelerating AI, robotics, and space ambitions as well, paving the way for orbital data centers.
The Tech Portal is published by Blue Box Media Private Limited. Our investors have no influence over our reporting. Read our full Ownership and Funding Disclosure â
