Meta has signed a 1GW capacity reservation agreement with Overview Energy for space based solar power, with an orbital demo planned for 2028 and commercial service targeting 2030. However, the best power transmission ever demonstrated in space—the Naval Research Lab's SWELL…
Meta has signed an agreement reserving up to one gigawatt of space-based solar power from a startup called Overview Energy. The capacity reservation, announced Monday, positions the deal as a first-of-its-kind agreement to beam solar energy from orbit to data centers. It is the kind of announcement that sounds like the future arriving on schedule.
The physics has some other ideas.
Overview Energy has no satellites in orbit. It has no operational hardware. Its planned in-space demonstration is scheduled for 2028 on a SpaceX Bandwagon-7 rideshare, with commercial service targeting 2030 at the earliest. In the meantime, the only laser power beaming experiment successfully operated in space produced 1.5 watts — not kilowatts, not megawatts. Watts. The Naval Research Laboratory's SWELL experiment, launched in 2023, transmitted that 1.5 watts across 1.45 meters of void at 11% end-to-end efficiency. To put that in context: it was enough to keep a dim flashlight running.
One gigawatt is one billion watts.
The gap between what Overview has contracted to deliver and what anyone has ever demonstrated transmitting through space is roughly nine orders of magnitude. That is not a refinement problem. That is a category difference.
The company is not unaware of this. Overview emerged from stealth in December 2025 with an aircraft demonstration over Pennsylvania, flying a Cessna Caravan at 15,000 feet in 70-knot crosswinds and validating that its optics could maintain pointing alignment with a ground receiver at 5 kilometers. That is a real result — acquiring and sustaining a laser lock on a moving target in turbulent conditions is genuinely hard. The company also says it validated its laser and optics at thousands of watts in the lab before that flight. Thousands of watts is a meaningful number. It is not one gigawatt, and it is not in orbit.
The Department of Defense has been working the same problem longer and with more resources. DARPA set a record in 2025 by transmitting 800 watts over more than eight kilometers. That test used stationary platforms on both ends. The moment either end of the link is moving — as a satellite in low Earth orbit necessarily is — the pointing and atmospheric compensation problem gets substantially harder.
Space-based solar power is not a new idea. NASA studied it in the 1970s and concluded the price tag, adjusted for inflation, would exceed one trillion dollars. Multiple companies and government programs have cycled through the concept since. Each time, the gap between the physics and the economics proved wider than the engineering.
What is different now, according to Overview, is the cost curve for solar panels, the availability of reusable launch, and a modular architecture that the company argues can scale without requiring a single trillion-dollar spacecraft. The company has Mike Griffin, a former NASA administrator, on its advisory board. Griffin, who has spent 48 years looking at this problem, said this is the first concept he has seen that he thinks might work.
That is a carefully worded endorsement from someone who has watched previous attempts fail. It is also not a demonstration.
Overview has raised roughly $20 million across two seed rounds. That buys a lab, a flight test, and enough engineering talent to publish results and book a rideshare slot. It does not buy a constellation of power-beaming satellites in geostationary orbit, which is what 1GW of commercial delivery would require. The capital needed to build that is measured in the billions, likely tens of billions, and the timeline runs to the mid-2030s at optimistic estimates.
Meta's 2024 data center consumption was above 18,000 gigawatt-hours, enough to power roughly 1.7 million American homes for a year. A one-gigawatt capacity reservation, if fully utilized, would cover a meaningful fraction of that load. But a capacity reservation is not a power purchase agreement. It does not necessarily commit Meta to any capital expenditure. It may simply mean Overview has priority access to Meta's demand queue if — and this is the operative word — the technology works, the satellites get built, and the economics close.
The 2028 LEO demo will be the first real test of whether the physics scales. If Overview can demonstrate multi-kilowatt transmission from orbit, the gap narrows from nine orders of magnitude to something that resembles an engineering roadmap. If it demonstrates the same 1.5 watts the NRL managed three years ago, the roadmap is not the problem.
This deal gives Overview a customer with deep pockets and a credible need. That is more than most space solar startups have had before. It does not change what has been demonstrated in orbit, what DARPA achieved with stationary hardware, or how far 1.5 watts is from one billion.