A $25 Billion Bet on AI Infrastructure from Elon Musk

Centered on 2-nanometer chip innovation, the planned plant could produce one terawatt of computing power annually, positioning it at the forefront of next-gen AI development.

Elon Musk has unveiled a significant development for the technology sector: Terafab. The CEO of Tesla announced plans for a $25 billion USD semiconductor factory in Austin, Texas, operating as a significant joint venture with Tesla, SpaceX, and xAI. Terafab aims to consolidate the whole chip manufacturing process—from original design to sophisticated packaging—within a single facility, focusing on the state-of-the-art 2-nanometer node. The facility is expected to produce an extraordinary 1 million wafer starts monthly at full capacity.

The facility will produce two primary products: inference chips to support Tesla’s Full Self-Driving system and Optimus humanoid robots, in addition to D3 chips specifically designed for the extreme conditions of space. Musk asserted that traditional external suppliers like as TSMC and Samsung are unable to meet the extensive scale required by his AI architecture. The most audacious aspect of the blueprint? Allocating 80% of Terafab's computational capacity directly into low Earth orbit. Musk contends that solar-powered data centers in orbit would ultimately surpass terrestrial servers, utilising continuous solar energy and the vacuum of space for natural cooling. Small-batch production is scheduled for 2026; nevertheless, industry experts are already highlighting significant financial and technological challenges, considering the trio's absence of conventional semiconductor fabrication expertise.