Social posts claim SpaceX is planning 300–500 GW orbital AGI servers

The basic story is real. SpaceX is pursuing orbital data centers for AI. But the specific viral claim people are sharing — that SpaceX is planning 300–500 gigawatts of orbital AGI servers — mashes together two different things: an actual FCC filing from January 30, 2026, and an older Elon Musk social post from November 2025 about what Starship might eventually deliver per year. (docs.fcc.gov) ### What did SpaceX actually file? SpaceX asked the FCC for authority to launch and operate a new “Orbital Data Center system” of up to 1 million satellites. The filing says these satellites would provide huge computing capacity for advanced AI models, run in low Earth orbit, and connect mainly through optical inter-satellite links rather than acting like ordinary intern(docs.fcc.gov)bruary 4, 2026. (docs.fcc.gov) ### Where did the 300–500 GW number come from? From Musk, not from the filing. In a November 19, 2025 X post, he wrote that Starship should be able to deliver “around 300 GW per year” of solar-powered AI satellites to orbit, “maybe 500 GW.” That is a throughput claim about how much solar-powered hardware Starship might place in orbit over time. It is not the same thing a(docs.fcc.gov)now. (finance.yahoo.com) ### Does the filing mention solar power and cooling? Yes, in broad strokes. The filing says the system would directly use near-constant solar power and pitch space as a place with very low operating and maintenance costs. It also leans on the usual orbital-compute argument — that waste heat can be rejected by radiating it into space instead of (finance.yahoo.com)tailed engineering plan proving 300–500 GW is near-term or economical. (cdn.geekwire.com) ### Why do people keep saying “AGI servers”? Mostly because that framing is catchy. The actual filing talks about advanced AI models and the applications that rely on them. It does not announce a dedicated “AGI server” program by that name. So the viral version adds a layer of interpretation on top of a real filing. (cdn.geekwire.com)ible? In pieces, yes. Solar power is stronger in orbit than on Earth’s surface, and radiative cooling is a real advantage. SpaceX also has an obvious edge if Starship eventually brings launch cost down hard enough. But the catch is brutal — mass to orbit, radiation tolerance, repairability, thermal design, networking, (cdn.geekwire.com)of the idea flags cost and scaling as the hard part. (convergedigest.com) ### So is this a hoax? No. There is a genuine FCC application, and it is unusually ambitious. What is misleading is the leap from “SpaceX filed for orbital AI data centers” to “SpaceX is imminently building 300–500 GW of orbital AGI servers.” The first part is documented. The second part is still a Musk-scale extrapolation. (docs.fcc.gov) ### Why does this matter now? Because it moves the idea out of pure sci-fi and into the regulatory system. SpaceX is no longer just hinting at space-based compute — it has put a formal application in front of the FCC. That does not make the giant power numbers real today, but it does mean the underlying project is real enough to file. (docs.fcc.gov)posts are overstating the scale and certainty, but they are not inventing the whole thing. SpaceX has really asked to build orbital AI data centers. The 300–500 GW figure is best read as Musk’s long-range ambition, not a confirmed deployment plan. (docs.fcc.gov)