Elon Musk Outlines Terafab Plan To Move AI Compute Growth To Space

TLDR

• Elon Musk said Terafab aims to produce one terawatt of AI compute each year.
• Tesla, SpaceX, and xAI are working together on the Terafab project.
• The factory would combine design, fabrication, testing, and redesign in one site.
• Musk said most future AI compute may run on solar-powered satellites in orbit.
• Starship is central to moving large volumes of AI hardware into space.

Elon Musk has unveiled Terafab, a proposed chip factory meant to expand AI compute on a much larger scale. The plan links semiconductor production with space infrastructure and future orbital computing systems.

According to the provided text, Tesla, SpaceX, and xAI are working together on the project. Musk said the goal is to produce one terawatt of computing each year, far above current AI chip output.

Terafab aims to unite the full chip process

Musk said Terafab would bring the full chip cycle into one facility. That includes mask creation, fabrication, testing, and redesign. The goal is to shorten development time and improve chip designs faster.

The first manufacturing site is expected to be built in Texas. The project would also receive support at the state level. Musk said current chip supply chains are too spread out and too slow for the output he wants.

He said, “We either build the Terafab or we don’t have the chips.” That statement reflects the scale of the project and the demand he expects in the coming years.

The facility is designed around two chip types. One type would support edge inference for products such as Tesla vehicles, Cybercab, and Optimus robots. The other would be made for space use under harsh thermal and radiation conditions.

Space based compute is central to the plan

Musk argued that Earth does not have enough power capacity for a terawatt of compute. The provided text says total US electricity generation is about 0.5 terawatts, which is below the project target.

Because of that limit, Musk said most compute would need to operate in orbit. He described solar-powered AI satellites as the main platform for that expansion. A prototype mini-satellite would produce 100 kilowatts.

Later versions would scale into the megawatt range. Reaching the full target would require launching about ten million tons of material into orbit each year. That estimate assumes 100 kilowatts of compute per ton.

Musk said, “Starship is a critical piece of the puzzle.” He tied the launch system directly to the wider plan for compute and power growth in space.

Starship and launch costs shape the roadmap

The current Starship V3 can carry about 100 tons to orbit per payload, based on the provided text. Musk said the future V4 version would increase that figure to 200 tons.

The same text says SpaceX has completed more than 500 booster landings. It also says launch costs have fallen from over $65,000 per kilogram in the Space Shuttle era to about $1,000 to $2,000 per kilogram today.

Musk said SpaceX wants to reduce that cost to $100 to $200 per kilogram. He believes that range could make space-based AI systems cheaper than ground systems within two to three years.

The provided text also says global AI compute capacity is about 20 gigawatts per year. Musk said existing suppliers remain important, but their expansion pace does not match his target. Terafab is presented as the response to that gap.

The post Elon Musk Outlines Terafab Plan To Move AI Compute Growth To Space appeared first on CoinCentral.