Elon Musk has announced a joint venture between two of his companies, Tesla and SpaceX, to build the world’s biggest computer chip facility.
The TeraFab project, which is expected to cost at least $20 billion, will aim to manufacture 50 times the number of AI chips that all of the world’s major companies currently produce annually.
Production will begin at the site of Tesla’s headquarters in Austin, Texas, with the facility set to produce two types of chips.
The first type of chip will be for Tesla’s self-driving vehicles, robotaxis, and its Optimus humanoid robots. The second type will be specially designed to withstand the harsh conditions of space.
Mr Musk has previously spoken of his ambition to create AI data centres in low-Earth orbit, citing the potential for lower costs and higher efficiency compared to ground-based setups.
“The lowest-cost place to put AI will be in space, and that will be true within two years, maybe three at the latest,” the tech billionaire said at the World Economic Forum in Davos earlier this year.
SpaceX recently filed plans with the US Federal Communications Commission (FCC) to establish a satellite data centre network.
Nvidia CEO Jensen Huang unveiled an initiative last week to help build AI data centres in space, while Amazon and Blue Origin founder Jeff Bezos has said orbital data centres could solve electricity demand constraints on Earth.
In announcing the new TeraFab project, Mr Musk said: “In order to understand the universe, you must explore the universe.
“And that’s the motivation to accelerate humanity’s future in understanding the universe and extending the light of consciousness to the stars.”
The tech boss has set out an aggressive timeline for TeraFab, with early production scheduled to begin next year, before mass production in 2028.
The two year timeframe is more ambitious than the industry standard, with semiconductor production typically taking three years to set up.
