The Orbital Mind
A story of AI data centers in space
I. The Heat Problem
Dr. Amara Okonkwo had spent her whole career fighting the same enemy: heat.
Every datacenter she'd ever built on Earth drowned in it. The GPUs ran hot, the cooling towers guzzled water, and the utility bills for the sprawling AI training clusters had grown so obscene that entire towns fought to keep the campuses out. By 2041, training a single frontier model consumed the annual electricity of a small nation — and dumped enough waste heat to warm a river.
"We're cooking the planet to build machines that tell us how to cool the planet," she muttered one night, staring at a thermal map glowing angry red.
That was the night she looked up.
II. Cold and Endless
Space, it turned out, was the perfect radiator.
Not because it was cold — vacuum has no temperature — but because it was empty. With nothing to conduct heat away, a datacenter in orbit could only shed warmth by radiating it into the infinite dark. Build the panels large enough, angle them away from the sun, and you had a cooling system that never ran dry and never sent a bill.
Enjoying the article?
Create a free account to finish reading and join the conversation.