Elon Musk’s planned Terafab semiconductor facility is shaping up to be an absolute monster. SpaceX and Tesla have now confirmed that the completed campus in Grimes County, Texas will feature more than 100 million square feet (9.3 million square metres) of manufacturing space. To put things in perspective, that’s roughly four times the combined floor area of Tesla’s Giga Texas, the Pentagon, Apple Park and the Mall of America. X user Nic Cruz Patane has also created a visualisation of the proposed site, offering a better idea of just how absurdly large Musk’s latest megaproject is.
The sheer size makes more sense when considering that Terafab is intended to be much more than a conventional semiconductor fabrication plant. Rather than simply producing AI processors, the facility is planned as a vertically integrated campus handling advanced logic and memory chip manufacturing, lithography, packaging and testing under one roof.
The announcement follows Tesla’s groundbreaking earlier this year on a research semiconductor facility at the North Campus of Giga Texas. That smaller facility is being positioned as a precursor to Terafab and will provide a starting point for Tesla’s in-house chip manufacturing efforts.
US$16.8 Billion To Start
SpaceX and Tesla said the initial phase of Terafab will require an investment of US$16.8 billion (~RM68.8 billion), with the completed campus expected to consist of four massive buildings. The companies have not yet detailed how these buildings will be divided between the various manufacturing and support operations, but the full project is expected to require substantially more investment as additional phases are built.
The facility is being developed to address what Musk considers a looming shortage of AI computing hardware for his companies. SpaceX and Tesla expect their combined semiconductor requirements to eventually exceed 1 terawatt (TW) of compute per year, a figure that is significantly beyond current global supply.
Terafab is expected to produce AI inference processors for Tesla’s Optimus humanoid robots and Cybercab autonomous vehicles, alongside high-power processors for SpaceX’s planned space-based data centres. The companies have not said when their combined demand will reach the 1 TW mark, however.
The project will also integrate Intel technology, with the initial phase expected to use the chipmaker’s 14A process. Tesla was previously announced as the first customer for the process, with the company planning to use it for future AI processors before Terafab eventually takes on high-volume production.
From Chip Fab To Entire Supply Chain
One of the more unusual aspects of Terafab is its attempt to bring multiple stages of semiconductor production together in a single location. Normally, logic and memory chips are manufactured at separate facilities using different processes, while packaging and testing are typically handled elsewhere.
SpaceX and Tesla believe consolidating these operations could shorten production cycles and speed up the time required to improve and refine new chip designs. It would also give Musk’s companies considerably more control over their semiconductor supply, rather than relying entirely on external chipmakers.
Musk first floated the idea of building his own chipmaking facility in late 2025, arguing that Tesla’s growing AI ambitions could eventually require more semiconductor capacity than the existing supply chain could provide. SpaceX and Tesla formally announced Terafab in March this year, with the project subsequently expanding into what could become one of the largest semiconductor manufacturing campuses ever proposed.
The scale of the ambition also comes with considerable financial risk. SpaceX has previously estimated that the Terafab programme could require up to US$119 billion (~RM486 billion) in investment, while acknowledging that there is no guarantee the ambitious project will ultimately succeed.
(Souce: SpaceX / Terafab / Nic Cruz Patane / Joe Tegtmeye)