AI Is Running Out of Room on Earth — So It’s Moving to Space

Here’s a constraint that doesn’t show up in any earnings call but is quietly strangling the AI build-out: Earth is running out of places to put data centers. Not chips. Not capital. Land, water, and power.

Gartner pegged 2025 AI infrastructure spending at $965 billion, with $1.37 trillion projected for 2026. McKinsey estimates $7 trillion in total data center investment will be needed by 2030. Yet Bloomberg reports that nearly half of all AI data center projects in the U.S. will be delayed this year — not because the money isn’t there, but because power grid interconnection queues now stretch three to five years. Microsoft, Alphabet, and Amazon have found themselves in the absurd position of holding the chips and the capital, yet unable to get a utility hookup in time. Water rights for cooling are drawing regulatory scrutiny across the American West. Prime real estate — cool climates, cheap power, fast fiber — is being consumed faster than it can be developed.

So where does AI go next? Orbit.

Orbital compute — AI data centers launched into low Earth orbit, powered by continuous solar energy, cooled by the near-vacuum of space, beaming results back via laser link — is no longer a futurist thought experiment. In November 2025, Starcloud launched a satellite containing an Nvidia H100 GPU, the first of its class in orbit. A month later it trained a language model in space — the first time that had ever been done. In January 2026, SpaceX filed with the FCC for authorization to launch up to one million satellites as orbital AI data centers. In March, Nvidia unveiled the Vera Rubin Space-1 module, a chip platform built specifically for orbital deployment. Jensen Huang called it “the final frontier.” And in April, SpaceX confidentially filed for an IPO targeting a $1.75 trillion valuation — the largest in market history — with orbital compute as the central thesis.

The economics look brutal today: running one H100-equivalent GPU for an hour costs roughly $1 on the ground; in orbit, it runs about $142. But $85 of that is pure launch cost. As SpaceX’s Starship drives launch prices toward $10 per kilogram — down from $2,000 per kilogram a decade ago — that premium collapses fast. The structural advantages of space (near-unlimited solar, free vacuum cooling, proximity to satellite-born data sources) become economically dominant before most people expect.

The investment angle here isn’t just SpaceX pre-IPO. It’s the entire supply chain: satellite hardware makers, space-grade solar panel manufacturers, laser communication companies, and anyone enabling edge AI at altitude. We’ve seen this playbook before — the early internet infrastructure buildout rewarded not just the household names but the picks-and-shovels players most investors never considered. The AI grid leaving Earth is the same story, one altitude higher. The constraint changes everything. The opportunity is just beginning to be priced in.