This article first appeared on GuruFocus .
Elon Musk says artificial-intelligence computing in orbit could grow faster than new data-center capacity on Earth within the next few years, setting up one of SpaceX's most ambitious expansion bets yet.
The SpaceX ( NASDAQ:SPCX ) CEO said orbital computing capacity could surpass new terrestrial installations by late 2028, arguing that terrestrial compute will be peanuts compared to orbital compute.
The claim builds on SpaceX's broader plan to push AI infrastructure into orbit using its launch network, upgraded satellites and increasingly powerful computing systems.
Musk's comments followed a financial model from Sequoia Capital partner Shaun Maguire that projects SpaceX revenue rising from $51.5 billion in 2026 to $274 billion in 2027 and $731 billion in 2028. Those are investor estimates rather than company guidance.
Maguire's model assumes Earth-based computing capacity reaches 21 gigawatts by the end of 2028, up from roughly 1 gigawatt in early 2026, while annual revenue per gigawatt approaches $40 billion. It also projects enormous capital needs, with 2028 capex potentially reaching $369 billionand free cash flow not turning positive until mid-2028.
SpaceX already has a large AI footprint on the ground through xAI.
Epoch AI estimates xAI's Colossus 2 facility in Memphis currently represents about 1.11 million Nvidia H100-equivalent GPUs, making it the world's largest AI computing site by that measure.
The company is also working toward orbital computing directly.
SpaceX is developing an AI satellite system around Nvidia's Vera Rubin platform and has said it expects to launch Nvidia-powered AI computers into orbit in 2027.
Investor Takeaway
I think the real question is not whether orbital AI sounds futuristic.
It is whether SpaceX can make compute in orbit economically better than adding another data center on Earth.
The theoretical advantages are obvious: abundant solar power, fewer terrestrial permitting constraints, and SpaceX controlling the launch infrastructure needed to deploy the hardware.
But the technical hurdles are equally real.
Cooling AI chips in space is difficult, radiation can damage hardware, and fast-moving chip cycles create the risk that expensive satellites become obsolete before their useful lives end.
For investors, the key things to watch are Starship launch costs, orbital power and cooling, Nvidia-powered satellite deployments, customer demand, and whether SpaceX can prove that orbital compute has better economics than terrestrial capacity.
Musk is not just talking about launching more satellites.
He is talking about moving a meaningful piece of the AI infrastructure industry off the planet.
