Intel confirmed it's partnering with Elon Musk's companies — Tesla, SpaceX, and xAI — on Terafab, a proposed semiconductor megacampus at the Giga Texas site in Austin. The stated goal: collapse the global semiconductor supply chain's distributed model (fab in Taiwan, memory in Korea, packaging in Southeast Asia) into a single 100 million square-foot facility capable of eventually producing one terawatt of computing power per year.
This is the marquee foundry customer Intel has been hunting since its IFS pivot. Getting Musk's combined AI and robotics demand as an anchor tenant is a genuine inflection point — if it executes. The colocation concept is genuinely interesting from an engineering standpoint: mask revision cycles that currently require intercontinental logistics could in theory collapse from weeks to days when litho, fab, and advanced packaging are on the same campus.
The detail buried in Lip-Bu Tan's statement — "step change in how silicon logic, memory, and packaging will get built" — is the advanced packaging angle. Intel's 3D packaging capabilities (Foveros, EMIB) are what differentiate it from TSMC's pure-play foundry model. Terafab may be the proving ground for whether Intel's integrated logic+packaging story can compete at the leading edge.
The caveat is the timescale. Building a 100M sqft fab complex from scratch in Austin, at the scale required to actually matter for Musk's demand, is a decade-plus project. The announcement is real; the terawatt number is aspirational. Watch for first-phase capacity targets and initial process node announcements — that's when this becomes more than a press release.