Intel officially joined the Terafab project on April 7, partnering with SpaceX, Tesla, and xAI on a chip manufacturing effort aimed at producing one terawatt per year of compute. [1] Intel will design, fabricate, and package ultra-high-performance chips at a facility in Austin, Texas, with CHIPS Act funding supporting the effort. Shares rose roughly four percent on the announcement. [2]

The partnership looks significant on paper. The question is whether Intel's foundry can deliver on it.

Elon Musk visited Intel just days before the announcement — a sequence that X read not as courtship but as audition. [1] Intel Foundry has spent years promising it would become a world-class contract manufacturer. It has not yet demonstrated that at scale. The company's process technology has lagged TSMC and Samsung for the better part of a decade, and no amount of government subsidy changes the physics of catching up. Terafab is simultaneously the largest vote of confidence the foundry has received and the highest-stakes test it has faced.

For Musk, the logic is vertical integration — the same instinct that built Tesla's battery factories and SpaceX's engine shop. Controlling the silicon supply chain for AI compute is the natural extension. [2] For Intel, the logic is survival. The foundry business needs anchor customers, and three Musk companies at once is as anchored as it gets.

Whether the silicon ships on time is the only question that matters. Everything else is announcement.

-- DAVID CHEN, Beijing