The data centers that power artificial intelligence have always sat on Earth, drawing huge amounts of electricity and water to keep servers running. Now the industry is looking to move that infrastructure to orbit, and the companies behind it include some of the biggest names in technology: Google, SpaceX, and a group of startups pushing the idea forward.
The proposed setup would sit in orbit and draw power from the sun. The move is being pitched as a way to free AI infrastructure from the constraints that have made terrestrial data centers controversial.
The Terrestrial Problem
Data centers are among the largest consumers of electricity on the planet. They run on a constant feed of power, and they need cooling systems to handle the heat they generate. That means they also consume large amounts of water, often drawn from local sources that can already be stressed.
The industry has tried to address this by building facilities in colder climates, where air conditioning runs more efficiently. Some centers have moved to coastal locations where seawater can be used for cooling. But those solutions only push the problem elsewhere. The demand for power and water remains, and so does the environmental footprint.
The result is a tension between the benefits of AI and the costs of running it. The technology has transformed everything from healthcare to climate research to everyday consumer services. But the physical plants that make it work are growing ever larger, and the planet is starting to feel the strain.
Why Orbit?
The idea of moving computing power into space is not new, but the current push appears to be reaching a critical mass. Companies are now openly exploring orbital data centers as a serious option, rather than a speculative idea.
The advantage is simple: the sun never sets in orbit. A satellite in the right position could capture solar energy continuously, without the daily cycle that limits ground-based solar farms. That steady stream of power could replace the fossil fuels and grid electricity that currently dominate the data center industry.
There is also a practical engineering argument. Space is cold. The vacuum of orbit offers natural cooling that terrestrial data centers must replicate through expensive equipment. A satellite could run its hardware at temperatures that keep it efficient without the need for chillers and fans.
The move would not eliminate data centers entirely. There will still be a need for edge computing and local processing, especially for applications that require low latency. But the core storage and heavy lifting could shift to orbit, where the environment is hostile but the energy source is abundant.
Who Is Involved
Google has been working on this area for years, and its investments in satellite technology suggest it sees a future where cloud computing extends beyond the planet’s surface. SpaceX is building Starlink, its constellation of internet satellites. Its expertise in orbital deployment and satellite operation makes it a natural partner for any company that wants to put hardware in orbit.
The startups involved are betting that the technical barriers to orbital computing can be overcome, and that the economics will eventually favor the move. The question is whether they can build a business case that competes with the entrenched interests of the terrestrial data center industry.
“The sun never sets in orbit.”
That line captures the core appeal of the proposal. In orbit, a satellite can harvest sunlight without interruption. The challenge is getting the hardware there, protecting it from radiation, and transmitting data back to Earth fast enough to matter.
The Constraints
The move is not without its challenges. Launching anything into orbit is expensive, and the cost of maintenance and repair in space is far higher than on Earth. Satellites fail, and when they do, they cannot simply be driven to a service center.
There is also the question of latency. Signals travel from a satellite to a user, and the distance involved adds time to requests. For applications that require real-time responsiveness, that delay could be unacceptable.
Then there is the regulatory question. Space is governed by international law, and launching a constellation of data centers raises questions about ownership, liability, and the environmental impact of space debris. The industry will need to navigate those issues before any orbital data center becomes operational.
The Big Picture
The move to orbit is part of a broader trend in technology. The cloud is becoming more distributed, with processing power spread across multiple locations rather than concentrated in a few giant facilities. The satellite companies are already changing how we connect to the internet, and the next step could be shifting computation itself off the planet.
The consequences of this move are significant. If it works, it could reduce the environmental footprint of AI while improving access to computing power around the world. The sun is a universal resource, and harvesting it in orbit could make computing available to regions that currently lack reliable power grids.
But there is also a risk. The concentration of power in a few large companies is already a concern, and giving those companies control over orbital infrastructure could amplify that problem.
The move to orbit is a bet on the future of computing. It is a bet that the sun is a better power source than coal or gas, and that the cold of space is a better cooling system than the air conditioning towers that currently define the data center landscape. Whether it pays off remains to be seen, but the companies making the bet are some of the most powerful in the world.
The move to orbit is a radical proposal, but it is not science fiction. The technology exists, the companies are committed, and the pressure to find a better way of running the machines that power our digital lives is only growing. The sun never sets in orbit, and that fact is now driving some of the most ambitious plans in the industry.
Where the paper stands
The paper backs narrow rules against direct harm from AI companies, including forcing disclosure of hidden safety failures, and is against federal licensing or pauses that freeze today’s leaders in place and lock out challengers. It is also against letting big companies hide their failures.
The orbit data center proposal carries the same risks as any other concentration of power, and the companies leading it are some of the most powerful in the world. The paper supports disclosure requirements so that safety failures are exposed, not buried, and opposes licensing schemes that only giants can afford.
What matters is keeping the field open to challengers, not freezing it in place.
Source material: “AI’s data center problem could move to space,” the Washington Examiner.
Get the Notebook.
The day's best stories and every fresh verdict, in plain English, in your inbox by seven. One email a day, no more.