Elon Musk claims orbital data centers could cost less than Earth-based ones within two to three years, but aerospace engineers say that timeline is unrealistic. SpaceX, which recently merged with Musk's AI company xAI, announced plans to launch data-crunching satellites into orbit to power artificial intelligence workloads. The pitch is simple: space is always sunny, power is free, and Earth's electrical grid is running out of capacity. Yet the obstacles are enormous, from cooling microchips in a vacuum to launching massive satellites at costs that remain prohibitively expensive .

Why Is Earth Running Out of Power for AI?

The numbers tell the story. Global data center power consumption is expected to roughly double to nearly 1,000 terawatt-hours by the end of the decade, according to the International Energy Agency . That's a staggering increase driven almost entirely by artificial intelligence workloads. Companies are already building dedicated gas turbines and investing in nuclear technology just to keep existing data centers running, yet it still isn't enough .

Philip Johnston, CEO and co-founder of Starcloud, a company actually building orbital data centers, painted a dire picture: "We're very quickly running up on constraints on where you can build new energy projects terrestrially. Within six months, they'll just be leaving chips in warehouses because they don't have power for turning them on," he stated . Starcloud launched its first spacecraft last fall with an Nvidia H100 chip on board and demonstrated the ability to run a version of Google's Gemini AI from space .

Google is pursuing its own orbital vision through Project Suncatcher, which envisions an 81-satellite cluster built in partnership with satellite-imagery company Planet. Two prototype satellites are scheduled to launch in early 2027 .

What Are the Engineering Challenges of Space Data Centers?

The engineering hurdles are formidable. To understand the scale of the problem, consider the International Space Station, currently the largest power-producing facility in orbit. Its solar panels, which span roughly half a football field, generate around 100 kilowatts of average power, according to Olivier de Weck, a professor of astronautics at MIT . To replicate a 100-megawatt data center in space would require a facility 500 to 1,000 times larger, depending on the orbit .

"Is that feasible? Yeah, I think it's feasible, but not next year and certainly not in three years," said Olivier de Weck, professor of astronautics at MIT.

Olivier de Weck, Professor of Astronautics, Massachusetts Institute of Technology

Power generation is only half the battle. Cooling is equally critical and far more complex in space. While Earth-based data centers use air conditioning, space offers no atmosphere to dissipate heat. Satellites get hot, and that heat has nowhere to go in a vacuum .

The solution involves massive radiators that move liquids to giant panels where heat can be dissipated. Combined with equally massive solar arrays, this means orbital data centers would need to be enormous or operate as sprawling satellite constellations . Rebekah Reed, a former NASA official now at Harvard University's Belfer Center for Science and International Affairs, explained the scale problem: "When you put those massive radiators together with massive solar arrays that are required in order to power and cool, you're actually talking about really large satellites, or very, very large satellite constellations," she noted .

How Would Orbital Data Centers Actually Work?

• Power Generation: Massive solar arrays spanning hundreds of meters would generate electricity, though current costs of around $1,000 per kilogram to launch satellites into orbit make deployment prohibitively expensive. Google estimates costs must drop to $200 per kilogram before orbital data centers make financial sense .
• Heat Dissipation: Radiator systems would move coolant to giant panels to shed heat in the vacuum of space, requiring engineering solutions that don't yet exist at scale .
• Data Transmission: Satellites would need to beam data between each other using lasers, but even at light speed, the latency between satellites is long enough to slow computing performance .
• Maintenance and Operations: Unlike Earth-based data centers staffed by workers, orbital facilities would require remote monitoring and autonomous systems to handle repairs and upgrades .

Google's Project Suncatcher proposes flying satellites in extremely tight clusters to reduce latency problems. Musk, meanwhile, has proposed launching upward of a million satellites and placing them in orbit around Earth's poles. He recently unveiled the first generation "AI Sat Mini" spacecraft with solar arrays spanning roughly 180 meters, or about 600 feet .

What's the Real Timeline for Orbital AI?

Musk's claim that orbital AI could cost less than terrestrial AI within two to three years is what Brandon Lucia, a professor of electrical and computer engineering at Carnegie Mellon University who specializes in putting computers on satellites, calls "an optimistic interpretation" . The napkin math looks appealing on paper, but reality is far messier.

Starcloud's second spacecraft, launching in October, will have 100 times the power generation of its first satellite, though it's still expected to generate only around 8 kilowatts of power . That's a tiny fraction of what a functional orbital data center would need. For context, a single terrestrial data center facility like DataBank's IAD1 in Ashburn, Virginia, consumes around 13 megawatts of power .

The cost of launching massive satellites remains the biggest barrier. At $1,000 per kilogram, launching the infrastructure needed for even a modest orbital data center would cost billions of dollars. Musk believes his Starship rocket, still in development, can eventually reduce launch costs dramatically. Starcloud's Johnston told investors: "If you don't think Starship's going to work, don't invest in us, that's totally fine," underscoring how dependent the entire orbital data center vision is on a single unproven technology .

Musk

Will Marshall, Planet's CEO, offered a more measured perspective: "Orbital data centers are an idea whose time has come. When exactly it will be more cost efficient than terrestrial ones is debatable but now is the time to be working on this," he wrote to NPR . That's a far cry from Musk's confident two to three year timeline.

Will Marshall, Planet's CEO

The race to orbit is real, and the engineering challenges are being tackled by serious companies with real prototypes in space. But the gap between prototype and practical, cost-effective orbital data center remains vast. For now, Earth's power crisis will have to be solved the old-fashioned way: more power plants, better efficiency, and perhaps a dose of nuclear energy .