Elon Musk says the future of computing lies entirely in orbit, a bold claim that landed just as Google revealed a plan to launch solar-powered AI satellites by early 2027.
The SpaceX CEO posted his prediction on X in response to a video circulating on the platform about Google's Project Suncatcher, an initiative that would place AI chips aboard satellites and use near-constant sunlight to power them. Musk's verdict was characteristically sweeping: the share of computing that eventually operates in space, he wrote, will "obviously round up to 100% of all compute."
That is not a hedge. It is a prediction that every server farm, every cloud operation, and every processing task now running on the ground will one day relocate above the atmosphere. Whether or not the timeline is realistic, the fact that the CEO of the world's dominant launch company is saying it publicly, while Google is already building prototypes, signals that major players in both tech and aerospace see orbital computing as more than a thought experiment.
Google's plan, as Fox News reported, centers on Project Suncatcher. The concept: launch satellites carrying AI chips into orbit, where they can tap into sunlight that is available almost around the clock. On Earth, solar power is interrupted by weather, nightfall, and atmospheric filtering. In orbit, those obstacles largely vanish.
Google's researchers described satellites flying in tight formation, linking their onboard chips to handle larger computing tasks, essentially creating a distributed data center in space. The company has partnered with Planet, a satellite imaging firm, to launch two prototype satellites by early 2027.
The project is still in its testing stage. No hardware has launched yet, and Google has not disclosed which specific AI chip models the satellites would carry or how much processing capacity the initial prototypes would deliver.
Adding confusion to the rollout, an unidentified X account that shared the Project Suncatcher video claimed Google would launch its chips aboard a SpaceX Falcon 9 rocket "next week." That claim has not been confirmed. Google's own published plan makes no mention of an imminent Falcon 9 launch, and neither Google nor SpaceX has publicly verified the timeline.
The gap between the social-media claim and Google's stated early-2027 target is wide enough to matter. Readers should treat the "next week" assertion as unverified.
Musk, for his part, did not address the Falcon 9 claim directly. His post focused on the broader trajectory, that computing will migrate off-planet, rather than on any specific launch schedule.
The logic behind orbital computing is straightforward, even if the engineering is not. AI chips are power-hungry. Training and running large models requires enormous electricity, and demand is growing faster than ground-based infrastructure can keep up. Data centers already strain local power grids in parts of Virginia, Texas, and Ireland.
Sunlight in orbit offers a partial answer. Without atmospheric interference, solar panels produce energy more consistently and at higher intensity. Google's pitch is that satellites could run AI workloads on solar power alone, sidestepping the land-use fights, permitting delays, and grid-capacity bottlenecks that slow new data center construction on the ground.
None of that means the obstacles are small. Launching hardware into space is expensive. Cooling electronics in a vacuum presents its own challenges. Latency, the time it takes for a signal to travel from orbit to a user on Earth, could limit which tasks orbital chips can handle. And maintaining or upgrading satellites is far harder than swapping components in a server rack in Iowa.
Musk's enthusiasm for space-based computing is not disinterested commentary. SpaceX operates the Falcon 9, the most frequently launched rocket on the planet, and runs Starlink, a constellation of thousands of satellites already providing broadband internet from low Earth orbit. If computing migrates to space at anything close to the scale Musk envisions, SpaceX would be positioned to launch, service, and connect much of that infrastructure.
Google, meanwhile, chose Planet, not SpaceX, as its named launch partner for the Suncatcher prototypes. Whether SpaceX ultimately provides launch services for the project remains an open question that neither company has addressed publicly.
Musk framed his prediction in absolute terms:
"The amount of compute in space will obviously round up to 100% of all compute."
The word "obviously" does a lot of lifting in that sentence. Ground-based data centers handle virtually all of the world's computing today. Moving even a fraction of that capacity into orbit would require a manufacturing, launch, and maintenance pipeline that does not yet exist. But Musk has built a career on stating the destination before the road is finished, and then building the road.
For now, Project Suncatcher remains a concept with a prototype target date. Google has not released detailed technical specifications, cost projections, or performance benchmarks. The two test satellites planned for early 2027 will be a first proof of concept, not a functioning orbital data center.
Several key questions remain unanswered. What role does Planet play beyond launch, does it provide satellite hardware, ground operations, or both? What computing tasks would the prototypes attempt? And how would Google handle the latency problem for real-time AI applications that require near-instant responses?
The gap between a bold social-media prediction and a working orbital computing network is measured in years, billions of dollars, and engineering problems that have not been solved yet. Musk and Google agree on the direction. The distance left to travel is the part neither one is spelling out.
When the two most powerful forces in tech and space both point the same way, it is worth paying attention, even if the destination is still a long ride from the launchpad.
