Google is preparing for a world where AI data centers become too large, too hot, and too power-hungry for Earth’s infrastructure.Project Suncatcher asks a stark question: if the next computing boom needs gigawatts of constant power, why not move the machines closer to the Sun?This documentary examines the economics behind orbital AI infrastructure: the substation bottlenecks slowing Microsoft, Amazon, Meta, and Google; the electricity demand that could push data centers past national-scale consumption by 2030; the water and cooling constraints reshaping site selection; and the brutal launch-cost math that still stands between one NVIDIA H100 in orbit and a true space-based cloud. The promise is continuous solar energy, fewer land battles, and machines that do not need rivers to stay alive. The risk is that the most important infrastructure of the AI age becomes controlled by the few companies capable of paying gravity’s price.CHAPTERS0:00 The Substation Wall0:39 Sunlight Above Earth4:13 The Orbital Ledger7:55 Machines Without Water13:43 One H100 in Orbit18:26 Everyone Pays the Toll25:32 Gravity’s Price31:38 The Next Frontier

Nvidia did not win the AI boom by accident — it turned itself into a toll road. By 2024, one company controlled the most important input in the industry: the chips that train models, with demand so intense that even $30,000 Blackwell systems were sold out before they shipped.Every major AI lab, from OpenAI to xAI, has been forced to pass through Nvidia’s gate, paying for access to the same scarce compute. This is the story of how a graphics chip maker became the backbone of the AI economy — and why that concentration of power is now shaping who can build the future, and how fast.

CHAPTERS

0:00 The Gate

0:35 The Scale

2:24 From Graphics to Gravity

4:12 The Mechanism

7:09 The Escalation

10:04 The Consequences

12:45 The Challenge

14:53 The Resolution