The proposed merger between Elon Musk’s SpaceX and xAI, exclusively reported by Reuters on Thursday, could give new impetus to Musk’s plans to launch satellite data centers into orbit as he battles for supremacy in a rapidly escalating AI race with tech giants such as Alphabet Inc.’s Google, Meta and OpenAI.
Here’s what we know about space-based AI computing.
What is a space-based AI data center?
Space-based data centers are still an early concept, but as energy-intensive facilities on Earth become increasingly expensive to operate, they will rely on hundreds of solar-powered satellites networked in orbit to handle the massive computing demands of AI systems like xAI’s Grok and OpenAI’s ChatGPT.
Proponents argue that operating above the atmosphere could provide near-constant solar power, eliminate the burden of cooling, which accounts for much of the cost of ground-based data centers, and make AI processing much more efficient.
But engineers and space experts warn that commercialization is still years away, citing significant risks from space debris, shielding the hardware from cosmic radiation, limited in-person maintenance options, and launch costs.
Deutsche Bank expects the deployment of the first small orbital data centers in 2027-2028 to test both technology and economics, and if early missions go well, wider constellations (with the potential to scale to hundreds or thousands) will emerge for the first time in the 2030s.
Why would Musk want to do this?
SpaceX is the most successful rocket maker in history, successfully launching thousands of satellites into orbit as part of its Starlink internet service.
If space-based AI computing is the future, SpaceX is ideally positioned to operate AI-enabled satellite clusters and facilitate the setup of on-orbit computing.
“It’s easy to build solar-powered data centers in space…The lowest cost place to deploy AI is going to be in space. That will be a reality within two years, at the latest within three years,” Musk said at the World Economic Forum in Davos earlier this month.
Reuters reports that SpaceX is considering an initial public offering this year that could value the rocket and satellite company at more than $1 trillion.
According to officials, a portion of the proceeds will be used to fund the development of an AI data center satellite.
What are Musk’s competitors doing?
Jeff Bezos’ Blue Origin is working on AI data center technology in space, based on the Amazon founder’s prediction that a “huge gigawatt data center” in orbit could outstrip the cost of a data center on Earth within 10 to 20 years by harnessing uninterrupted solar power and radiating heat directly into space.
Nvidia-backed Starcloud is already showing a glimpse of that future. The company’s Starcloud-1 satellite, launched last month on a Falcon 9, is equipped with the Nvidia H100, the most powerful AI chip ever placed in orbit, to train and run Google’s open source Gemma model as a proof of concept.
The company ultimately envisions a modular satellite “hypercluster” that would provide about 5 gigawatts of computing power, comparable to multiple hyperscale data centers combined.
Google is advancing the idea of space-based data centers with Project Suncatcher, a research effort that will network solar-powered satellites equipped with Tensor Processing Units into an in-orbit AI cloud.
The company plans to launch an early prototype with partner Planet Labs around 2027.
China also plans to launch space-based artificial intelligence data centers and build a “space cloud” over the next five years, state media reported on Thursday.
China’s main space contractor, China Aerospace Science and Technology Corporation, has vowed to “build gigawatt-class space digital intelligence infrastructure” in accordance with a five-year development plan.
