Elon Musk has a habit of describing his own projects in absolute terms, but even by that standard, Thursday's announcement stood out. Terafab, he wrote on X, will be the largest and most valuable building on Earth by far. On August 6, 2026, Tesla and SpaceX confirmed they're jointly investing $16.8 billion in an initial phase to build it, a semiconductor complex planned for more than 100 million square feet in Grimes County, Texas, roughly an hour northwest of Houston.

The site sits next to the Gibbons Creek Reservoir, which once cooled a coal plant that shut down in 2018 and will now supply water for the new fab instead. The companies say the facility will employ at least 3,000 people, drawn largely from Grimes and neighboring Brazos County.

The Biggest Building on Earth, Built to Solve One Problem

Terafab isn't a vanity project dressed up in Musk's usual grandiose language, even if the language is doing its usual work. According to SpaceX's own announcement, the facility exists specifically to bridge the divide between current global chip supply and the compute demand of the future, a gap Musk's companies say they're already running into. Tesla and SpaceX collectively project needing more than one terawatt of computing power annually in the coming years, a figure that, according to the companies, exceeds current and projected global chip manufacturing capacity entirely, not just what Musk's businesses alone could realistically buy from existing suppliers.

That's the real motivation behind building a chip fab rather than simply placing bigger orders with existing manufacturers. If the global industry genuinely can't produce enough advanced chips to meet demand this steep, buying more aggressively from the same limited supply doesn't solve the problem. Building new capacity does, at least in theory, assuming the fab actually gets built at the scale being promised.

What Makes Terafab Different From a Normal Chip Fab

Most semiconductor facilities specialize narrowly, focusing on either logic chips or memory chips, and typically outsourcing packaging and testing to separate specialized companies entirely. Terafab is designed to break that pattern, housing logic chip production, memory chip production, packaging, and testing all under one roof. Musk has described targeting 2-nanometer process technology, placing the facility's ambitions at the leading edge of what's currently manufacturable anywhere in the world.

The output is split between two very different use cases. One line of chips is destined for Tesla's Optimus humanoid robots and its Cybercab robotaxis, hardware that needs to run sophisticated AI models locally and reliably. The other is built specifically to withstand the extreme conditions of space, feeding SpaceX's ambitions around orbital data centers, an idea the company has floated as an eventual complement to its ground-based compute infrastructure. Bringing both product lines into a single vertically integrated facility is an unusually aggressive bet on internal chip independence, at a moment when even the largest technology companies still rely heavily on outside foundries for their most advanced silicon.

The Numbers Behind How Big This Could Actually Get

The $16.8 billion figure covers only the initial phase. A filing SpaceX submitted back in May put the company's own proposed investment at $55 billion for the base build, with that number climbing to as much as $119 billion if every planned expansion phase gets completed. Whether Terafab ultimately reaches that scale depends on a lot of variables still playing out, including whether Tesla and SpaceX's actual chip demand grows the way Musk is projecting.

Intel's Role Is Real, But How Real Is Unclear

Intel signed on earlier this year to help Tesla and SpaceX refactor their silicon fab technology, lending its own manufacturing expertise to a project being led by companies with no prior experience running an advanced chip fab at this scale. That partnership adds real credibility to Terafab's technical ambitions, since Intel remains one of the few companies in the world with genuine institutional knowledge of building and operating leading-edge fabs, expertise it's also been trying to monetize through its own foundry business with outside customers like Fortinet.

What's notable is how quiet SpaceX stayed about Intel's involvement during its own first earnings call as a public company earlier this same week. Terafab didn't come up at all in that call, according to TechCrunch's reporting, an odd omission for a project this size involving one of SpaceX's most significant technology partnerships. That silence doesn't necessarily signal trouble, but it's a gap worth watching as more concrete details about Intel's actual role and financial stake in the project emerge over time.

Not Everyone in Grimes County Is Celebrating

Local reaction hasn't been uniformly positive. A heavily attended county meeting held the day before Thursday's formal announcement drew hundreds of residents raising concerns about the scale of tax breaks awarded to the project and what they described as a lack of transparency in how the deal came together. Texas Governor Greg Abbott, for his part, confirmed a $30 million Texas Enterprise Fund grant supporting the project and designated Terafab a qualified development under the state's Jobs, Energy, Technology, and Innovation program, framing it as a major economic win for the region.

That tension, state and local officials celebrating an enormous jobs and investment announcement while some residents question the terms attached to it, is a familiar pattern for large industrial developments, and it's one that's likely to keep surfacing as construction actually begins.

A Bet That Fits a Bigger Pattern

Terafab lands in the middle of a broader scramble across the tech industry for chip capacity that already touches Meta's own decision to build custom silicon, Samsung's relocation of its US operations to the same state Musk is now building in, and Intel's push to land outside customers for its foundry business. What sets Musk's approach apart is the sheer scale of vertical integration he's attempting: rather than partnering more closely with an existing chipmaker, Tesla and SpaceX are trying to build the entire supply chain themselves, betting that owning the fab outright is worth the enormous upfront cost and execution risk. Whether that bet pays off depends on questions Thursday's announcement didn't answer: how quickly Terafab can actually reach production at the scale Musk described, and whether Tesla and SpaceX's compute needs grow fast enough to justify a $119 billion commitment before global chip supply catches up on its own.