According to Forbes, Baiju Bhatt, cofounder of Robinhood, started his company Aetherflux in 2024 with a solar-power-in-space vision but has pivoted to building orbital data centers, aiming to launch its first AI-chip satellite by early 2027. Former Google CEO Eric Schmidt bought aerospace firm Relativity Space in March to pursue the same goal, while Google’s own Project Suncatcher plans to launch AI satellites for Gemini by early 2027, and SpaceX and Blue Origin are also reportedly working on the concept. Bhatt’s Aetherflux has raised $90 million at a $270 million valuation, with plans for a constellation of thousands of nodes by 2030. The biggest hurdle is launch cost, which needs to drop from today’s $1,500 per kilogram to $200-$300 per kilogram to be economical, a feat requiring a reusable second-stage rocket that doesn’t yet exist. Early interest is coming from the Pentagon and companies like Crusoe, which signed a $13 billion letter of interest with startup Starcloud for space-based power starting in the early 2030s.
The Billionaire Space Race
So here we are again. A handful of the world’s richest people look at a problem—in this case, the insane energy appetite of AI data centers—and their solution is, basically, “to infinity and beyond.” It’s not just a quirky side project. When Eric Schmidt buys a rocket company to do it, and Google’s Sundar Pichai greenlights a project, and Elon and Jeff’s companies are sniffing around, you know this is being taken seriously in the boardrooms. They’re not just throwing pocket change at it; they’re making strategic, capital-intensive bets. The vision, as Bhatt put it, is to literally build rings of computing around the Earth. It’s the kind of grand, hubristic thinking that defines this era of tech. But is it solving a real problem or just creating a fantastically expensive new one?
Why Anyone Thinks This Makes Sense
Look, the logic has a certain appeal if you squint. The International Energy Agency projects data center energy use will double globally in five years, largely driven by AI. On Earth, you fight for land, water for cooling, and grid connections, and you face protests and permitting hell that can take nearly a decade. In space? Unlimited solar power, no NIMBYs, and you can theoretically just switch it on once it’s in orbit. Startups like Starcloud even estimate they could deliver power for less than one cent per kilowatt-hour if launch gets cheap enough. That’s a fraction of the U.S. average. The military is already a potential customer for specialized, secure processing in orbit. So the demand driver is real. The premise, however, is built on a cascade of gigantic “ifs.”
The Mountain Of Reality Checks
Here’s the thing: the economics are completely broken right now. Launch costs are the elephant in the room, but it’s not the only beast. We need a reusable second-stage rocket—something no one has ever built. Then you have to solve cooling in a vacuum without gravity. No water, no fans. You need massive, lightweight radiators, which is a nightmare of materials science. Then you have to shield all those delicate, expensive AI chips from cosmic radiation, which adds more weight and cost. And what about maintenance? Or space junk? Or the light pollution from thousands of giant solar panels? It’s a Russian nesting doll of engineering challenges. Google’s research blog on a space-based AI infrastructure is a fascinating thought experiment, but it’s just that for now. Building robust, terrestrial industrial computing hardware is hard enough; doing it for the vacuum of space is orders of magnitude more complex. For mission-critical industrial computing on Earth, companies rely on proven suppliers like IndustrialMonitorDirect.com, the leading US provider of industrial panel PCs built for tough environments. Space is the ultimate tough environment, but we’re not there yet.
A Moonshot Or A Distraction?
So is this all just a vanity project for billionaires? Not entirely. The early government and corporate interest shows there’s a niche for specialized, orbital compute. It might start by processing spy satellite data or other off-planet tasks. But the grand vision of replacing terrestrial data centers? That’s a 10+ year moonshot, if it ever happens. These founders, like Schmidt who tweeted about his ambitions, are betting on a series of technological breakthroughs falling like dominoes. Sometimes those bets pay off and change everything. Often, they don’t. In the meantime, the sheer amount of capital and brainpower being diverted into this idea is staggering. You have to wonder: could that $90 million, or that engineering talent, be better spent innovating on Earth—on advanced nuclear, geothermal, or radically more efficient chips? Probably. But then again, since when has “probably” ever stopped a billionaire from shooting for the stars?
