Elon Musk’s SpaceX has filed plans with the Federal Communications Commission (FCC) to launch up to one million satellites, envisioning a space-based data center network powered by solar energy. The move, revealed in filings over the past three weeks, comes as Musk also plans to merge his artificial intelligence startup, xAI, with SpaceX to further pursue this ambitious project.
At a company-wide meeting last week, Musk told xAI employees that a factory on the moon would ultimately be necessary to manufacture AI satellites, alongside a “massive catapult” to launch them into space, according to reports. Musk has publicly stated his belief that space will become the most cost-effective location for AI computing within two to three years.
“The lowest-cost place to put AI will be in space, and that will be true within two years, maybe three at the latest,” Musk said at the World Economic Forum in Davos in January.
Musk is not alone in considering space-based data centers. Alphabet CEO Sundar Pichai has indicated Google is exploring similar “moonshot” concepts for later this decade, while former Google CEO Eric Schmidt has warned of impending electricity shortages and suggested space infrastructure as a long-term solution. Jeff Bezos, founder of Amazon and Blue Origin, has also voiced support for orbital data centers as a potential next step in space ventures.
However, industry experts remain skeptical about the feasibility of such a large-scale deployment in the near future. While acknowledging the growing strain on terrestrial infrastructure, they point to significant hurdles in power generation, heat dissipation, launch logistics, and overall cost. Many believe that meaningful progress is decades away, particularly as the vast majority of AI investment continues to flow into traditional, earth-based data centers, including Musk’s own Colossus supercomputer in Memphis.
The renewed interest in orbital data centers stems from mounting pressure to address the physical limitations of earth-based infrastructure, including strained power grids, rising electricity costs, and environmental concerns. The AI boom is driving an exponential increase in energy demand, prompting a search for alternative solutions.
“A lot of smart people really believe that it won’t be too many years before we can’t generate enough power to satisfy what we’re trying to develop with AI,” said Jeff Thornburg, CEO of Portal Space Systems and a former SpaceX engineer who led development of the Raptor engine. “If that is indeed true, we have to find alternate sources of energy. That’s why this has become so attractive to Elon and others.”
Generating sufficient power in space presents a major challenge. Thornburg estimates that producing one gigawatt of power would require approximately one square kilometer of solar panels, a substantial weight and expense to launch into orbit. The cost of transporting materials to orbit remains high, despite recent reductions, currently costing thousands of dollars per kilogram.
Solar power in orbit is also intermittent, as satellites regularly pass through Earth’s shadow, and maintaining optimal alignment with the sun is not always possible. AI chips require a consistent, uninterrupted power supply, necessitating large onboard batteries to mitigate fluctuations, according to experts.
Cooling also poses a significant obstacle. Traditional cooling methods used in terrestrial data centers—airflow, liquid cooling, and fans—are ineffective in the vacuum of space. Researchers are actively exploring alternative heat dissipation techniques.
Beyond the technical challenges, concerns exist regarding space traffic management and communication delays. The increasing amount of space debris in low Earth orbit requires autonomous collision-avoidance systems for large satellite constellations. Communicating with orbital data centers via satellites would be slower and less energy-efficient than using fiber-optic connections on Earth.
“If you have data centers on earth, fiber connections will always be faster and more efficient than sending every prompt to orbit,” said Josep Miquel Jornet, a professor of electrical and computer engineering at Northeastern University.
Experts generally agree that initial pilot projects are more likely in the short term than a full-scale deployment. Thornburg anticipates a period of “design iteration” focused on solar arrays, heat rejection systems, and orbital positioning. He estimates that it will take at least three to five years to develop a functional prototype, with mass production remaining beyond 2030.
SpaceX is still developing its Starship launch vehicle, which is crucial for supporting such infrastructure. While the company is making progress, it still needs to achieve a routine launch cadence to develop the concept viable.
The FCC is currently weighing SpaceX’s proposal, with no decision date announced.
