This, of course, is about Elon Musk’s rocket company SpaceX, which he believes can become an AI player after essentially inventing the modern space business and creating the first killer space application, satellite broadband, enabled by lower launch costs.
SpaceX has had quite a month. Its private market valuation recently hit $800 billion, based on reports of a secondary sale that provided insiders with some liquidity, making it more valuable than OpenAI or TikTok parent ByteDance. SpaceX didn’t respond to Barron’s request to comment on its valuation. Then Musk all but confirmed plans for a 2026 IPO at a rumored $1.5 trillion valuation.
His rocket company, however, is profitable, according to Musk, and doesn’t have a pressing need for cash. Why bother with an IPO then?
There is a new killer application that SpaceX can dominate if it moves quickly. AI data centers in space. That’s right, it sounds like science fiction, but AI computers in orbit powered by the sun might be a lower-cost way of getting answers from chatbots or training cars to drive themselves.
Speed is of the essence. OpenAI, Jeff Bezos’ Blue Origin, and Alphabet are looking at ways to do the same thing, wrote Deutsche Bank analyst Edison Yu in a Monday report. (There’s nothing like an arms race to motivate Wall Street.)
SpaceX, however, is holding a pair of aces.
For starters, it is a near-monopoly provider of space launch services. It accounts for more than half of all orbital launches annually. Excluding China’s launches and SpaceX’s share is closer to 70%, and the tonnage it can deliver to space exceeds that figure. No one can put satellites in space—for AI, wireless communications, or defense applications—faster than SpaceX.
Ace number two is Musk. He also runs Tesla, which is putting AI into machines. And xAI, which owns X, is competing with OpenAI’s ChatGPT and Alphabet’s Gemini.
“With AI driving up demand for compute, Earth-based data centers appear to be encountering certain structural bottlenecks, mainly energy, cooling, and latency,” added Yu. But space is cold, the sun can always shine at the right orbit, and lasers in a vacuum can even outperform fiber-optic cables.
To be sure, there will be challenges around cost and maintenance. But the idea is viable, says Mark Boggett, investment manager at Seraphim Space, a U.K.-based firm investing in early-stage space companies.
“I liken it to where we were with the [telecom] market…three years ago nobody was taking [satellite to mobile phone] seriously,” he said.
Now, space-based broadband is capable of delivering phone and data services on par with terrestrial infrastructure. And space-based defense applications in the U.S. and Europe are proliferating. Think, President Donald Trump’s proposed Golden Dome missile shield.
Satellite technology has improved, but the biggest innovation was reusable rockets, which dramatically lowered the cost to reach space. It cost tens of thousands of dollars to put a kilogram of weight into orbit on the Space Shuttle. SpaceX’s Falcon Heavy rocket can do it for thousands of dollars, or less. SpaceX’s Starship rocket system, being tested, is designed to do it for hundreds of dollars.
Yu said three companies could benefit as space data center hype grows. Rocket Lab has been described as a mini-SpaceX, providing launch and satellite services. Planet Labs is working with Alphabet to develop satellites. And Intuitive Machines is buying satellite company Lanteris. “We think Intuitive Machines will gain access to a [communication] platform which can be used to house [AI chip] payloads.”
It’s too early to call out space AI winners and losers. Those three are as good a place to start as any. So is EchoStar, which sold SpaceX wireless spectrum for a price including $11.1 billion worth of SpaceX stock when the rocket company was valued at closer to $400 billion.
