Elon Musk's plans for SpaceX data centers in orbit could cost $5 trillion a year, analysts say

By William Gavin

Analysts at research firm MoffettNathanson say 'the capital needs would be simply enormous'

SpaceX CEO Elon Musk wants to launch a constellation of 1 million satellites into orbit.

The tie-up of SpaceX and xAI is all about a plan to dominate space with artificial-intelligence data centers, according to Elon Musk, who founded both companies. It won't be cheap.

Analysts at MoffettNathanson crunched the numbers and concluded "the capital needs would be simply enormous" and require "a staggering amount of external financing." In a note published on Tuesday, they found that SpaceX's plan to power AI could cost as much as one-sixth of the current $31 trillion gross domestic product of the United States, depending on how the company pursues its plan.

MoffettNathanson analysts led by Nick Del Deo broke down what would be needed to reach Musk's goals, beginning with a potential 1 million-satellite constellation that SpaceX is seeking government approval to build. On the high end of the MoffettNathanson projections, SpaceX could be shelling out $5 trillion on capital expenditures each year on the effort.

To start, SpaceX would need to launch about 200,000 satellites into orbit each year, assuming that each satellite can last five years before being replaced, the analysts said. Based on details SpaceX has shared about the weight of the next-generation Starlink satellites and Starship's launch capacity, the Del Deo's team expects it will take about 3,300 launches a year, or nine per day, to meet Musk's goals.

If each Starship can be reused 100 times, at least 30 rockets would need to be built each year, along with additional vehicles for other missions, the analysts said. That would also require "radically expanded" supply chains to support such a massive undertaking, they added.

Musk on Monday wrote that the "basic math" behind his goals would have SpaceX add 100 gigawatts of AI computing capacity each year. The MoffettNathanson analysts estimate that 100 gigawatts of orbital compute "might" require capital expenditures of between $4 trillion and $5 trillion if SpaceX buys equipment from Nvidia (NVDA).

The analysts estimate an Nvidia-based equipment deployment could cost between $40 billion and $50 billion per gigawatt in that scenario. SpaceX's spending could be lowered to the tune of $2 trillion if the company uses custom silicon, Del Deo's team said.

Read: SpaceX officially acquires xAI. Here's how Elon Musk justifies the move.

SpaceX will need to find someone with deep pockets to finance Musk's vision. The company brought in an estimated $15 billion of revenue in 2025 and is projected to generate $23.8 billion in 2026, according to a forecast from Payload, a space news and research publication.

XAI won't be of much help on that front. It spent $7.8 billion in cash across the first nine months of 2025 and reported a $1.46 billion loss for the September quarter, according to Bloomberg News. Much of its money has been allocated toward building out infrastructure.

Some of the cash SpaceX needs to advance its plans is likely to come from an initial public offering, which is reportedly targeted for mid-June. The company is looking to raise up to $50 billion at a valuation of as much as $1.5 trillion, according to the Financial Times.

Read: The 'space economy' is booming. Here's where most of the money is going.

There are also a number of other issues that SpaceX needs to overcome. Deutsche Bank analysts have noted that space radiation can accelerate the degradation of chips and in-orbit maintenance can be "impractical," among other things.

Musk and other proponents think orbital data centers can solve an artificial-intelligence power issue. While SpaceX is the only company currently pursuing a plan of that magnitude, it's not alone in pursuing space-based computing. At least six other U.S. companies, including Alphabet (GOOG) (GOOGL), also have plans for orbital data centers.

"Global electricity demand for AI simply cannot be met with terrestrial solutions, even in the near term, without imposing hardship on communities and the environment," Musk said in a memo published Monday. "In the long term, space-based AI is obviously the only way to scale."

A lunar focus

Although Musk has called lunar missions a "distraction" in the past, he's softened his stance. Mars, the planet he hopes to colonize, was mentioned just twice in a statement announcing that SpaceX had acquired xAI, while the moon got four mentions.

Part of that comes down to NASA's move to open up the contract previously awarded to SpaceX for the Artemis III and Artemis IV missions. Both SpaceX and Amazon (AMZN) founder Jeff Bezos' Blue Origin have been asked to submit plans to help accelerate the development of a lander for the moon missions.

"We are absolutely moving forward with both acceleration paths in parallel," NASA Administrator Jared Isaacman told the New York Times last week.

In his memo to staff, Musk said Starship will be capable of landing "massive amounts" of cargo on the moon, enabling the permanent establishment of manufacturing and research facilities. Factories on the moon, he added, could make satellites and deploy them further into space.

"The capabilities we unlock by making space-based data centers a reality will fund and enable self-growing bases on the Moon, an entire civilization on Mars and ultimately expansion to the universe," Musk said.

Don't miss: Now that SpaceX and xAI have merged, is Tesla next?

-William Gavin

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

02-04-26 0939ET