Tesla’s Terafab readies launch
Tesla plans to bring its $20B “Terafab” AI chip factory online within days to produce what the company says could be hundreds of billions of AI chips annually, targeting robotaxis, Optimus projects and edge robotics compute reported. The vertical integration push aims to optimize silicon specifically for real‑time robotics inference — a big signal that system‑level hardware/software co‑design is becoming competitive strategy. This raises pressure on legacy foundries and shifts the talent map toward edge AI and hardware/software codesign skills.
Elon Musk posted on X that the “Terafab Project launches in 7 days,” pointing to March 21, 2026. (teslarati.com) Tesla’s public coverage ties the facility to a roughly $20 billion price tag. (teslarati.com) Separate reporting projects an annual output between about 100 billion and 200 billion AI/memory chips, figures repeated across multiple industry write‑ups. (creati.ai) Other trade outlets characterize the scale in wafer terms, citing plans variously described as ~100,000 wafer starts per month up to as high as 1 million wafer starts per month in different briefings. (techreport.com) Tesla’s next‑gen AI5 chip is already slated to be manufactured at Samsung’s Taylor, Texas facility and at TSMC’s Arizona site, with company comments indicating volume vehicle integration won’t occur until around mid‑2027. (thetechportal.com) Industry leaders have pushed back on the timeline and difficulty—Nvidia CEO Jensen Huang called building frontier fab capability “extremely hard,” while reporting has noted Tesla has been in exploratory talks with Intel about supply and partnership options. (tomshardware.com) Recruiting activity reflects the pivot to silicon: Tesla posted AI chip design openings in Korea and is advertising process and chip roles in Austin and Palo Alto, and public job listings for hardware/software co‑design engineers have proliferated across leading AI organizations. (cnbctv18.com)
