Musk’s Company Wants to Do What Only NVIDIA and TSMC Have Managed — SpaceX Announces Its Own Chip Production

Ahead of a summer IPO that could value SpaceX at $1.75 trillion, the company has filed an S-1 registration with the SEC. Buried among the usual sections on risks and expenses was a surprise: SpaceX plans to produce its own GPUs.

The reason is simple — there aren’t enough chips. According to Elon Musk, even buying up all available chips on the market would cover only about 2% of the company’s future needs. SpaceX openly warned investors: it lacks long-term supply agreements with many chip vendors and has no guarantees it can secure enough hardware to sustain growth.

The proposed solution is Terafab — a joint chip manufacturing initiative between SpaceX, xAI, and Tesla in Austin, Texas. Musk has stated that the project will focus on chips for vehicles, humanoid robots, and orbital data centers. Manufacturing is expected to involve Intel and its next-generation 14A process.

The problem is the lack of specifics — and there are many unknowns. It’s unclear what kind of chips Terafab will actually produce: full-scale GPUs like those from NVIDIA, or specialized AI accelerators similar to Google’s TPUs. There’s no timeline, and no clear structure for who will handle what within the project.

And these aren’t minor details. Advanced chip manufacturing is one of the most complex processes in the tech world: thousands of tightly controlled steps, exotic materials, and near-atomic precision. TSMC spent decades and billions to reach its current position. Even NVIDIA designs chips but outsources production to TSMC. SpaceX, for now, is not a semiconductor company — though it may be trying to become one.

Do you think Musk can realistically challenge NVIDIA and TSMC — or is this too ambitious even for him? Share your thoughts in the comments.