Elon Musk has introduced a vision to create Terafab, which is set to become the most significant semiconductor manufacturing facility ever. This venture is a joint effort between Tesla, SpaceX, and xAI, revealed during a livestream on Saturday as a move towards establishing a “galactic civilization.” The facility, projected to cost between $20 and $25 billion, aspires to have a far-reaching influence that goes beyond Earth.
At first, Terafab will be a large chip manufacturing facility located next to Tesla’s Giga Texas in Austin. Musk pointed out that the plant will facilitate the production, testing, and enhancement of chips all in one place, a capability that is not present anywhere else globally. The ambition is for Terafab to ultimately generate 1 million chip wafers per month using 2-nanometer process technology, as mentioned at the Tesla shareholder meeting in November of the previous year.
Presently, Taiwan’s TSMC stands as the largest semiconductor producer, aiming to reach a monthly output of 140,000 2-nanometer chip wafers by the close of 2026. Musk foresees Terafab producing 100 to 200 billion AI and memory chips every year, chiefly for Tesla’s deployment in its vehicles and robots. He remarked that existing suppliers such as Samsung, TSMC, and Micron face constraints in expansion capabilities.
Musk underlined the critical need for Terafab, asserting, “That rate is much less than we would like. And so, we either build the Terafab, or we don’t have the chips.” Despite the audacious nature of this project, particularly for companies inexperienced in semiconductor manufacturing, Musk’s vision extends even further.
When fully operational, Terafab aims to provide 100 to 200 gigawatts of computing power on Earth and 1 terawatt in space, with 80% of its processing output based on SpaceX’s solar-powered AI satellites. Space presents advantages such as greater solar irradiance and simplified cooling, enhancing chip performance.
Currently, Earth’s low orbit has limited computing capacity, primarily restricted to satellite onboard processing. China’s “Three-Body Computing Constellation” serves as a notable case, with ambitions to significantly escalate its AI model capability. However, Musk’s aspirations are considerably more ambitious.
TSMC’s preeminence in semiconductor manufacturing is a crucial geopolitical concern, as the U.S. worries about China’s potential dominance over Taiwan. Establishing a large-scale data center in space poses its own obstacles, but SpaceX’s cutting-edge aerospace proficiency offers a distinct competitive edge.
Musk’s overarching vision encompasses the development of a “mass driver” on the moon to elevate computing output from terawatts to petawatts, although specifics remain unclear. The timeline for these advancements is unpredictable. The AI5 chip is expected to go into volume production in 2027, but the construction of Terafab will likely span several years. Musk’s previous ambitious plans, such as those presented at Tesla Battery Day in 2020, have not all come to fruition, highlighting the challenges that lie ahead.