Elon Musk's large-scale chip manufacturing plan bets on space computing power competition

Source: Economic Information Daily

American entrepreneur Elon Musk recently announced the launch of a chip production plan named “Terafab,” to meet the massive market demand arising from the linkage between the artificial intelligence and space exploration industries, with its planned production scale expected to exceed the combined total of the current chip production giants.

According to a report by Agence France-Presse, Musk said that Terafab’s manufacturing facilities will be located near Austin, Texas, with the goal of producing chips with the computational capability equivalent to 1TW every year.

Musk also said that the project will be jointly advanced through collaboration between his electric vehicle company Tesla and his rocket company Space Exploration Technologies Corporation (SpaceX).

He did not disclose the amount of initial investment for the project. However, according to earlier reports by U.S. media, the investment scale is expected to be between $20 billion and $25 billion.

Although Musk does not have experience related to semiconductor manufacturing, he said that the reason he is moving forward with the Terafab project is that Tesla and SpaceX’s expected demand for computational capability is expected to far exceed the production capacity of global chip suppliers.

“We are very grateful for the existing supply chain, grateful to companies like Samsung and TSMC, and we hope they can expand production capacity as soon as possible. We will purchase all the chips they produce. I have expressed this idea to them, but there is a limit to how quickly they can expand capacity, and this pace is far lower than our expectations, so we need to build Terafab,” Musk said.

Musk said that the “advanced technology semiconductor manufacturing plant” newly built in Austin will have the capability to design, manufacture, test, and improve every single chip.

Musk previously took a high-profile initiative to apply artificial intelligence technology to the space sector, driving leapfrog development in the space exploration industry, and announced that SpaceX would acquire his AI company xAI. In this transaction, SpaceX is valued at $1 trillion, and xAI is valued at $250 billion.

Citing statements from supporters of the aforementioned M&A deal, Reuters reported that the acquisition of xAI will help SpaceX advance its plan to build data centers in space. With the development of artificial intelligence driving a surge in demand for computing power, Musk believes that space-based facilities have higher energy efficiency than facilities on Earth. As for how Musk will obtain the enormous funds needed to build data centers in space, according to a prior report by Bloomberg News in the U.S., SpaceX is considering an initial public offering (IPO) as early as June this year, with a fundraising size of up to $50 billion. Also, according to The Wall Street Journal, SpaceX is expected to create the largest IPO in the history of the U.S. stock market.

Industry views hold that Musk is building his “Starlink AI” layout. SpaceX has filed an application with the U.S. Federal Communications Commission, proposing to deploy a system composed of 1 million satellites in low Earth orbit to build an in-orbit data center network, meeting high-performance computing needs such as those for artificial intelligence.

“Where it’s cheapest to deploy artificial intelligence is in space.” Musk said not long ago while attending the World Economic Forum Annual Meeting 2026 in Davos, Switzerland. He also claims that the deployment of artificial intelligence data centers in space “within two years, at most three years” will become a reality.

A commentary by CNN said that this acquisition and merger both indicates that xAI has a need for funding in the rapidly growing artificial intelligence field, and also highlights the importance of artificial intelligence technology for future space exploration.

In addition, the latest released Terafab plan shows that the project’s goal is to produce chips that can support 100GW to 200GW of computing power on Earth each year, as well as chips that can support 1TW of computing power in space.