SpaceX and Tesla confirm Texas site for Terafab, a major new AI chip manufacturing complex with Intel and xAI as partners—the first large-scale domestic foundry capacity dedicated to AI accelerators.

SpaceX and Tesla have formally confirmed the site for Terafab, an enormous artificial intelligence chip manufacturing complex being developed in partnership with Intel and xAI. The confirmation followed a $10 million payment SpaceX sent to Grimes County, Texas, earlier this week, satisfying a deadline built into a tax abatement agreement signed in June. Foundation preparation is expected to begin almost immediately, with construction renderings expected within days and physical construction to follow within months.

SpaceX selected the Gibbons Creek Reservoir, the former home of the Gibbons Creek Steam Electric Station, as the location for the full-scale facility. The company cited the site's rainwater storage capacity and its proximity to Houston's labor pool, with more than 15.9 million people living within a three-hour radius of the property. The complex will be powered by SpaceX's own natural gas plants rather than drawing electricity from Texas's ERCOT grid. SpaceX has acquired the Navasota River pumping station to divert stormwater into the reservoir for cooling and industrial use.

Terafab brings together three companies tied to Elon Musk—SpaceX, Tesla and xAI—alongside chipmaker Intel, which joined the project in April following a meeting between Musk and Intel CEO Lip-Bu Tan at Intel's campus. Musk first unveiled the venture in March at the defunct Seaholm Power Plant in Austin, describing it as "the most epic chip-building exercise in history." An initial prototype fabrication facility is already underway at Tesla's Giga Texas campus in Austin, designed to feed into the larger, full-scale Grimes County complex.

SpaceX has estimated that the first phase of the Grimes County site could cost approximately $55 billion, with total costs across all planned phases potentially reaching $119 billion. At full scale, Terafab is targeting 1 terawatt of annual AI compute capacity—roughly double current total U.S. semiconductor output. The project plans to reach 1 million wafer starts per month using Intel's advanced 14A manufacturing process, representing something on the order of 50 times current worldwide production of roughly 20 gigawatts annually.

Intel stated that its ability to design, fabricate and package ultra-high-performance chips at scale would help accelerate Terafab's aim of producing 1 terawatt per year of compute to power advances in AI and robotics. Intel CEO Lip-Bu Tan characterized the partnership as representing a step change in how silicon logic, memory and packaging will be built in the future.

The scale of the undertaking traces back to supply constraints SpaceX has described in its regulatory filings. In its Form S-1 filed ahead of its recent initial public offering, SpaceX warned that its orbital AI ambitions depend on chip access significantly beyond what is currently available, pointing specifically to capacity constraints at established foundries including TSMC and Samsung Foundry. Musk has repeatedly argued that existing global chip fab output covers only a small fraction of what Tesla and SpaceX will eventually require to power vehicles, Optimus humanoid robots and space-based data centers.

According to Musk's stated division of labor, Tesla is focused on the smaller prototype fab, while SpaceX is taking the lead on the initial phase of the full-scale Grimes County facility. The venture is designed around two primary facilities: one focused on terrestrial AI chip production for Full Self-Driving software and the Optimus robot, and a second, more experimental facility intended to produce chips specifically engineered for orbital data centers.

Some industry analysts have offered a more measured view of what Terafab represents, arguing it functions less like an independent Tesla-built chip fab and more like an Intel Foundry expansion project for which Tesla, SpaceX and xAI are serving primarily as anchor customers. That analysis noted that leading-edge chip fabrication typically takes established players like TSMC, Samsung and Intel roughly a decade and tens of billions of dollars, making it considerably more difficult for an automaker, rocket company and AI startup to independently build a competitive advanced process node.

SpaceX's regulatory disclosures have echoed this caution. The company's Form S-1 explicitly warns investors that Terafab "may not be successful" in closing the chip supply gap it was designed to address. Some industry analysts have separately calculated that achieving the full 1-terawatt annual capacity target would require capital expenditures ranging from $5 trillion to $13 trillion over the project's full lifespan, given the roughly 22.4 million advanced logic wafers such output would require processing each year. To help generate additional liquidity for its broader ambitions, SpaceX previously acquired xAI in an all-stock merger, creating a combined entity valued at approximately $1.25 trillion.

With site confirmation now finalized and construction expected to begin within months, Terafab moves into a phase where its backers' ambitious timelines will face their first real test. Traditional semiconductor fabs typically take five to seven years to progress from groundbreaking to full volume production, meaning even an aggressive construction schedule would likely place meaningful chip output several years away, even as the underlying chip supply constraints the project is meant to address continue to intensify across the technology industry.