$55 billion! Musk's "chip factory" is becoming a reality

In the 1980s, Japan’s semiconductor industry was at its peak, once occupying more than 50% of the global chip market share.

The Americans couldn’t sit still— not because they couldn’t buy chips, but because they realized that if a country couldn’t control its own “silicon,” it was essentially handing over its strategic lifeline to others. Thus came the “Semiconductor Agreement,” the later revitalization of Intel, and today’s “CHIPS Act.”

Forty years later, similar anxiety appears in a more personal way— in Elon Musk himself.

Only this time, the one who wants to control “silicon” isn’t a country, but a single person— and his cluster of companies.

On May 6, Bloomberg disclosed a document: SpaceX has formally proposed to invest $55 billion in Grimes County, Texas, initiating the "Terafab" chip manufacturing facility construction plan. If subsequent phases are all implemented, total capital expenditure may reach $119 billion.

Let’s convert those numbers to get a sense of the scale.

$55 billion is over twice TSMC’s total 2023 revenue. $119 billion is close to Nvidia’s revenue peak for the whole of fiscal year 2024. This isn’t just an “investment,” it’s an all-in bet— or, rather, a strategic declaration.

This project is a joint plan of SpaceX and Tesla. Musk’s logic is clear: his group of companies—SpaceX, Tesla, xAI—consumes an astronomical amount of GPU computing power every year. Training Grok needs chips, Starlink’s ground station network needs chips, Tesla’s autonomous driving needs chips, and future humanoid robots Optimus will also need chips.

Rather than handing money to Nvidia every year, why not just pay himself?

Strategically, this logic is flawless.

01

Musk’s Ambition for Vertical Integration

To understand Terafab, you first have to know what Musk has been doing these past two years.

In 2025, xAI acquired social media platform X. Earlier this year, SpaceX swallowed xAI via an all-stock deal. At the same time, SpaceX’s IPO plans are underway, with the roadshow window reportedly set to open around June 8, and the S-1 registration document expected to be filed later this month.

This is an increasingly dense web: rockets, satellite broadband, AI models, social platforms, electric cars, humanoid robots… and now, chip manufacturing is to be part of the territory.

DataCenter Knowledge’s analysis gets to the point: Terafab should not be understood simply as a “factory,” but as a “full-stack AI infrastructure strategy”— aiming to unify compute production, energy procurement, and deployment of computing resources under one roof.

This is a bit like how Amazon didn’t just want to sell books but also built its own freight network, data centers, and logistics satellites—except Musk is adding one more thing: a factory to produce the computers themselves.

When Apple started in-house development of its A-series chips, it was considered one of the most successful vertical integration decisions in tech history. But Apple only “designs” chips; manufacturing is still TSMC’s job. What Musk wants, is to take on manufacturing as well.

Even Apple didn’t dare to touch this ambition.

02

The “15-Year Strategy” Behind Everything

However, there is always a wide chasm between strategic vision and engineering reality.

Ben Bajarin, a chip analyst at Creative Strategies, used an interesting phrase: he says Musk is executing a “15-year strategy”—it sounds complimentary, but the implication is: don’t expect to see returns anytime soon.

Morgan Stanley’s forecast is even more direct. Their estimate is that even under the most optimistic construction assumptions, Terafab’s initial chip output will not be available until mid-2028 at the earliest. From today, that’s still more than two years away. By then, who knows what generation AI chip technology will have reached.

The brutality of semiconductor manufacturing is that it is the least forgiving industry for “PowerPoint promises.”

Building an advanced process wafer fab typically takes 3-5 years, requires extremely sophisticated lithography machines (only ASML worldwide can make high-end EUV), thousands of highly specialized chemical engineers, a stable supply of ultrapure water and power, and a full set of precision cleanroom management systems. Intel has spent tens of billions of dollars and years, but still struggles to catch up to TSMC in process technology.

Finance Monthly’s analysis points out this risk directly: chip projects are easily underestimated for their execution difficulties, they are slow, expensive, prone to delays, dependent on specialized machinery, skilled labor, and an already extremely tight supply chain.

Interestingly, when Intel CEO Pat Gelsinger was asked about Terafab, he gave a subtly worded response, saying he was “excited to explore innovative ways to reshape silicon process technology.” This can be read as a signal for cooperation, or as an indirect acknowledgment of supply-demand tensions in the market—or both.

03

More Than Just Business

But if you look at Terafab simply as an investment for business ROI, you might be applying the wrong analytical framework from the start.

The really interesting aspect is how it reflects the entire AI industry’s growing obsession with “computing autonomy.”

In the past three years, the essence of the AI arms race has shifted from “whose model is smarter” to “who can secure more computing power.” Nvidia’s H100, H200, GB200 chips are perpetually out of stock, TSMC’s advanced process manufacturing capacity is locked up years in advance. Microsoft, Google, Amazon, and Meta have all invested tens of billions of dollars in custom AI chip design and production.

Musk’s logic is essentially the same as these tech giants: In the AI era, computing power is a means of production— who controls the chips, controls AI.

And with Terafab’s announcement coinciding with SpaceX’s upcoming IPO, there’s another layer of meaning. As one tech news editor commented, part of this announcement is designed to “tie embattled Tesla together with the soon-to-be-listed SpaceX and the AI supercomputing narrative.”

This judgment may not be entirely fair, but it’s not unfounded. Just a few months ago, Musk personally admitted that “xAI was built the wrong way,” necessitating intervention from SpaceX to integrate it. Against this backdrop, Terafab’s high-profile debut is both a strategic layout and a capital story— these two are not contradictory.

Musk never does just one thing at a time.

Right now, the launch window for Starship Flight 12 is between May 12 and 18, and the Dragon cargo spacecraft will also fly to the International Space Station on the same day. SpaceX’s rocket business is still roaring ahead.

Yet in Grimes County, Texas, a piece of land that could rewrite the AI-era chip supply chain landscape is still just a coordinate on a piece of paper.

Whether $119 billion can materialize into an actual operating fab, no one can guarantee. But one thing is certain: when a company known for building rockets decides to build chips, yet again, this industry’s boundaries are being redefined.