Space based solar power has had no shortage of white papers. What it has never had is a customer. That changed this year, and the reason is not orbital mechanics. It is data centers.
The Meta agreement
Overview Energy and Meta announced a first of its kind agreement giving Meta early access to up to a gigawatt of capacity from Overview’s space solar system. Initial orbital demonstration is slated for 2028, with commercial power delivery targeted for 2030.
The architecture is the interesting part. Overview does not beam power to a new purpose-built receiver farm. It collects sunlight in orbit and beams it down onto solar facilities that already exist on the ground. Those panels then produce electricity at night and through cloud cover, which raises the utilization of assets a utility has already paid for. That is a considerably easier sell than building a rectenna field from scratch.
Virtus Solis and the 500 megawatt line
Separately, Virtus Solis confirmed it has signed terms for an initial deployment contract at 500 megawatt scale, with pilot operations targeted inside 24 months. The company is led by a former SpaceX engineer, which matters mostly because launch cost is the entire economic argument here and he knows exactly what it costs.
Why AI load changed the math
Space solar has always had one unbeatable selling point and one fatal flaw. The selling point is that a station in the right orbit sees the sun roughly 99 percent of the time, with no night, no weather and no seasonal loss. A two gigawatt station could in principle run two million homes continuously.
The fatal flaw was cost per kilowatt hour against terrestrial solar plus batteries, which kept falling faster than anything in orbit could keep up with.
AI training load broke that comparison. Hyperscalers are now buying against a constraint that is not price, it is availability of firm, always on power in a grid that cannot deliver it fast enough. When the alternative is not getting the electricity at all, a premium per kilowatt hour stops being disqualifying.
The part nobody has solved
Getting a gigawatt down from orbit means megawatt-class wireless transmission over roughly 36,000 kilometers. The best terrestrial numbers this year are 1,180 watts over 100 meters and just over a kilowatt by laser at short range. Caltech’s MAPLE experiment transmitted from orbit to Earth at milliwatt levels.
The gap between a demonstration and a deployment is six orders of magnitude. Signed contracts do not close it. They just mean somebody is now paying to find out.