Elon Musk has revealed Terafab, a hyper-scale chip manufacturing manufacturing facility aimed toward unlocking large AI compute and in the end supporting space-based infrastructure and human growth past Earth.
Tesla, SpaceX, and xAI, now a part of Musk’s aerospace firm, are collectively growing the venture. Their aim is to supply one terawatt of computing annually, roughly 50 instances the present world output of AI chips.
“The best way to truly scale civilization is to scale energy in area…as a result of we truly seize such a tiny quantity of the solar’s vitality on Earth,” Musk stated in a latest SpaceX broadcast.
“We need to be a civilization that expands to the galaxy with spaceships that anybody can go anyplace they need at any time,” he added. “To try this, we have to harness the ability of the solar. A Terafab, whereas it’s huge, a terawatt of compute per 12 months is gigantic by our kind of civilizational requirements, it’s nonetheless only one step alongside the way in which of being even a Kardashev.”
The fabrication facility
Terafab would combine the total chip improvement cycle right into a single facility, based on Musk. The system would come with capabilities for lithography masks creation, chip fabrication, testing, and redesign, enabling a speedy suggestions loop to iterate on chip designs.
Musk instructed that this method may significantly speed up enchancment cycles in comparison with the fragmented construction of at present’s chip provide chain.
The venture is predicted to start with a complicated manufacturing facility in Texas, supported by state-level backing.
Two chip classes
The initiative envisions two distinct chip classes. The primary can be optimized for edge inference, the type of onboard processing required by Tesla’s Optimus humanoid robots and its autonomous car fleet, together with the forthcoming Cybercab.
Musk tasks that humanoid robotic manufacturing may ultimately attain one billion to 10 billion models yearly, dwarfing the roughly 100 million automobiles produced worldwide annually.
The second chip variant can be purpose-built for area situations, engineered to face up to high-energy particle bombardment, and designed to run at elevated temperatures to cut back the mass of thermal radiators on orbiting platforms.
Why area, not Earth
Musk argued that terrestrial energy constraints make it bodily not possible to deploy a terawatt of compute on Earth, the place complete US electrical energy era hovers round 0.5 terawatts. As a substitute, the majority of computing infrastructure would orbit the planet aboard solar-powered AI satellites.
A prototype “mini-satellite” specification requires 100 kilowatts of output, with future iterations scaling into the megawatt vary. Attaining the total terawatt goal would require launching roughly ten million tons of fabric to orbit annually at an effectivity of 100 kilowatts per ton.
The present Starship V3 variant can ship roughly 100 tons to orbit per payload, a determine that the forthcoming V4 model would double to 200 tons. SpaceX has accomplished greater than 500 profitable booster landings and pushed launch prices from over $65,000 per kilogram throughout the Area Shuttle period to an estimated $1,000 to $2,000 per kilogram at present.
The corporate’s acknowledged ambition is to push that determine to between $100 and $200 per kilogram with Starship optimization, a threshold Musk believes would make space-based AI deployment cheaper than ground-based alternate options inside two to a few years.
To assist the transition, Musk pointed to the event of absolutely reusable launch techniques equivalent to Starship, which he stated will likely be important for transporting the large volumes of apparatus required.
“Starship is a important piece of the puzzle, as a result of with the intention to scale compute and scale energy, it’s important to go to area, which signifies that you want large payload to area. And Starship will allow that,” Musk stated.
He additionally outlined long-term ideas, together with using lunar-based manufacturing and mass drivers to additional scale back the price of deploying infrastructure into orbit.
The size of the hole
International AI compute capability is roughly 20 gigawatts per 12 months. Each semiconductor fabrication plant on the planet mixed accounts for roughly 2% of what Terafab would wish to achieve its one-terawatt annual goal.
Musk emphasised that current suppliers, together with main world foundries, stay important, however famous that their most snug growth price falls far in need of his necessities.
“We both construct the Terafab or we don’t have the chips,” he stated. “And we’d like the chips, so we construct the Terafab.”
