$50 million. That's what Baiju Bhatt, the Robinhood co-founder, raised from Index Ventures, Andreessen Horowitz, and Bill Gates's Breakthrough Energy Ventures to turn a 1941 Isaac Asimov short story into America's first orbital power grid. His startup, Aetherflux, plans to launch a satellite next year that will collect solar energy in low Earth orbit and beam it to the ground using infrared lasers.
The demo is set for 2026. If it works, the company's follow-up, a constellation called "Galactic Brain," will skip the terrestrial grid entirely: solar panels in orbit, compute nodes in orbit, power delivery via laser to military outposts and disaster zones. First commercial data center node: Q1 2027.
Galactic Brain orbital data center targets Q1 2027, powered by continuous space solar
DoD awarded proof-of-concept funding for laser power beaming to remote and contested environments
The window that finally opened
Space-based solar power (SBSP) has been a "five years away" technology since the 1970s. The physics was never the problem. The economics was. Geostationary platforms the size of cities with microwave rectennas measured in square kilometers needed launch costs that didn't exist.
What changed is not the architecture. It's the price of getting mass to orbit. SpaceX's Starship is pushing cost toward $200 per kilogram, a 95% reduction from the Space Shuttle era. At that number, the financial barrier that killed every previous SBSP program collapses.
A solar panel in geostationary orbit sees the sun 99% of the year. No clouds. No night. No atmosphere. The capacity factor jumps from 20% to 25% for the best terrestrial sites to above 95%. The output is baseload, the kind of power grids pay a premium for.
Aetherflux is betting the architecture should look nothing like the 1970s blueprints. Instead of one giant satellite in GEO, it's launching many small satellites in LEO. Instead of microwave beams requiring giant rectennas, it's using infrared lasers that need only a portable ground station five to ten meters across.
$60M total capital raised ($10M founder + $50M Series A)
$200/kg Starship-driven launch cost enabling the new math
95% orbital solar capacity factor vs 25% on the best terrestrial sites
5 to 10m portable ground station diameter
Galactic Brain: why AI compute needs orbit
In December 2025, Aetherflux announced a shift that turned a space-energy story into an AI-infrastructure story. "Galactic Brain" is an orbital data center, a satellite node that uses continuous solar power and radiative cooling to run GPU-class compute workloads in space. First commercial node: Q1 2027.
AI compute demand is outstripping terrestrial power infrastructure. A single data center can consume 500MW. Grid interconnection timelines run five to seven years. Building in orbit bypasses land acquisition, grid connection, and cooling water. The sun provides power, space provides cooling, and optical inter-satellite links provide connectivity.
"The race for artificial general intelligence is fundamentally a race for compute capacity, and by extension, energy," Bhatt told SpaceNews. "The elephant in the room is that our current energy plans simply won't get us there fast enough. Galactic Brain puts the sunlight next to the silicon and skips the power grid entirely."
Aetherflux's first mission, a subscale satellite built on Apex Space's Aeries bus launching via SpaceX Falcon 9 rideshare in 2026, will demonstrate laser power beaming from LEO to a ground station. The same laser and precision-tracking technologies are the building blocks for the orbital data center. Power beaming and orbital compute are, as the company puts it, "two sides of the same coin."
The customer that changes the calculus
Aetherflux has a customer most clean-energy startups don't: the U.S. Department of Defense. Through the Operational Energy Capability Improvement Fund (OECIF), the DoD awarded funding for a proof of concept demonstrating power transmission from LEO. The military spends $16 billion annually on energy, much of it on fuel convoys that put troops at risk.
"Space solar power can give the US an asymmetric advantage by delivering energy where it's needed most, day or night," Bhatt wrote on X. "This capability is especially valuable in remote or contested areas, including the Indo-Pacific, where traditional power delivery can be dangerous, expensive, or challenging."
The defense angle de-risks the business model. Aetherflux doesn't need grid-competitive pricing to generate revenue in its first phase. It needs to be cheaper than a diesel convoy in a conflict zone. That's a lower bar, and it buys time for the technology to mature toward civilian grid applications.
The field: who else is building in orbit
Aetherflux is not alone. A wave of startups is racing to commercialize space-based power, each with a different bet on the right architecture.
Star Catcher Industries, based in Jacksonville, Florida, raised $88 million (Series A led by B Capital) to build an orbital power grid that beams energy to satellites rather than Earth. Its approach uses optical power beaming to existing satellite solar arrays and targets the in-orbit market first. The company set the world record for optical power beaming and completed an on-orbit subsystem demonstration in 2025. Former U.S. Space Force chief General John W. "Jay" Raymond joined its board.
Space Solar, a UK-based startup, completed its Harrier 360° wireless power transmission demonstrator and raised £1.7 million from the UK Space Agency. Its CASSIOPeiA satellite concept targets GEO-scale power delivery. Reflect Orbital raised $6.5 million for orbital mirrors that redirect sunlight to Earth. The concept is simpler but limited to daylight hours.
Virtus Solis and the institutional programs — Caltech's SSPD-1, JAXA's OHISAMA, ESA's SOLARIS — round out a field that has more technical validation than commercial revenue. The trigger mechanism across all of them is the same: launch cost collapsed, and nobody has proven the business model yet.
What happens to space solar power between now and 2030?
Probability: 65%. Launch costs continue to fall, DoD procurement provides anchor demand, and orbital data center pilot programs validate the compute-in-space model. But scaling from kilowatt demonstrations to megawatt-class delivery will take longer than current roadmaps project. The gap between demo and revenue is three to five years, not one to two.
✅ Arguments for
— DoD procurement creates a non-grid-competitive revenue channel that funds iterative development
— AI energy demand is so acute that terrestrial buildout cannot keep pace, pushing compute operators to consider orbital alternatives
Confirmation criteria: Aetherflux or Star Catcher achieves orbital demo within the next 18 months with measurable power transmission
❌ Arguments against
— Regulatory and spectrum coordination for orbital laser power beaming is unproven. Airspace integration alone could take years
— No startup in this space has demonstrated kilowatt-scale orbital-to-ground power transmission yet; credibility rests on lab demos and paper designs
Disconfirmation criteria: Terrestrial energy storage costs fall below $50/kWh before any SBSP demo reaches orbit, erasing the premium that justifies orbital power
Aetherflux 2026 demo mission: successful LEO-to-ground laser power transmission
Star Catcher orbital power node deployment: first in-orbit energy delivery to a client satellite
Galactic Brain Q1 2027 target: does the first commercial orbital data center node launch on schedule?
DoD OECIF follow-on awards: scale of military procurement commitment beyond proof of concept
Development scenarios
🟢 Optimistic scenario (20%)
Implications: The space solar market skips the valley of death entirely and enters a rapid scaling phase. Aetherflux and Star Catcher become the new SpaceX and Iridium of energy.
🟡 Base-case scenario (55%)
Implications: Space solar power becomes a real but modest industry, think a few billion in annual revenue by 2032, not the trillion-dollar market proponents project.
🔴 Pessimistic scenario (25%)
Implications: SBSP becomes a cautionary tale in space-tech investing, and the trillion-dollar market thesis collapses into a footnote. The next credible attempt waits for another order-of-magnitude launch cost reduction or a geopolitical energy shock.
