Putting data centers in space has gone from science fiction to commercialization. SpaceX, Meta and Google all follow this concept, and for good reasons: the orbit provides near continuous solar energy and the natural cooling of the area, which prevents energy, allows and cools the obstacles that currently suppress the construction of a data center on Earth. In April, Meta announced that it had already appeared put a gigawatt future solar orbital capacity. As AI’s demand for electricity exceeds what terrestrial grids can comfortably provide, the commercial concept is hard to ignore.
But the argument is stuck on the wrong question. We’re still debating whether orbital data centers are technically feasible and how cheaply we can launch them—a debate even industry leaders continue to have. In February, OpenAI’s Sam Altman called for rotating data centers “It’s funny now,” calculates high failure rates and costs. The most important question is what happens when something goes wrong—and where the real risk lies when it does. It’s easy to think that risk is always in orbit. Mostly, it doesn’t. It sits on the ground, in several stations that connect the satellites back to Earth. Today, those facilities support the entire orbital computer system, yet receive only a fraction of the defense of their strategic importance.
In the world, a hardware failure is a technician entering the server room. In orbit, a simple glitch or micro-meteorite strike can sideline the equipment until the next launch window, turning what would otherwise be routine maintenance into months of lost capacity. This is not imaginary. In March, the SpaceX Starlink satellite had a mysterious anomaly on the routeThe type of error is that the ground is a repair ticket and in orbit it can cut off the material for good. The technology to operate servers in space is advancing faster than the technology to service them once they arrive. Until in-orbit maintenance becomes effective, every failure is a loss, and that resets the entire business case. Average repair time is no longer measured in hours. It is measured in the launch windows.
That fact should redefine the system design. If failure cannot be prevented in the way it happens on the ground, it should be expected instead. Redundancy and good degradation cease to be good things and become the whole structure of buildings, distributing workloads to many satellites and orbiting planes so that the loss of one platform reduces capacity instead of disabling operations.
The trap is a collective space: when several tenants share a single orbital platform, a single physical effect becomes a shared, irreversible termination. And the economy is unforgiving. Radiation, hot cycling and exposure to collisions turn common mistakes into total losses, driving insurance and redundancy costs far beyond what many energy-savings think. A business that uses one or two satellite heavy loads is sitting on one critical point of failure that no startup costing spreadsheet can adequately capture.
But even the in-orbit case is part of the problem. All orbital data centers are still dependent on Earth, substations, downlink centers and fiber networks carry data its last mile to users. A satellite is only as resilient as the ground infrastructure that connects it to the network, as well as those resources it is much easier to disrupt than a solid terrestrial data center. Focus on global computing behind a few channels, and create attractive bottlenecks for anyone who wants to cause interference, whether it’s congestion, signal interference or physical interference. Computer power may remain in orbit. The most accessible risk remains firmly on the downside.
This is the part of the discussion that gets the least attention if it is to be argued that it deserves the most. Like computer orbital measurements, these are low resources become the new crown jewels of the digital economy. The unfortunate reality is that the stability of a space-based data center will depend more on the development of its satellites than on the security of an otherwise unremarkable structure on Earth.
Protecting that infrastructure is not a cybersecurity problem, and treating it as one is a mistake. Satellites, ground stations and terrestrial networks operate as one interconnected system, meaning that compromises anywhere can permeate the entire architecture. A breached ground station is simultaneously a physical security incident and a cyber event.
Therefore, protecting these sites means abandoning the practice of disconnected security measures—a camera here, an access reader there, a visitor log on a clipboard—and moving toward integrated security structures where physical access, video, alarms and proprietary systems work together. Organizations will need analytics that can identify anomalies before they become incidents, rather than simply documenting what has already happened. As these resources be a single point of failure in the orbital compute, that consistency becomes the difference between a contained event and an outage on a national scale.
None of this is an argument against putting a computer in space. The power case is real, and the companies that follow it are serious. It’s an argument for honestly risking prices before we get there, for investing in service, redundancy and safety for a lower class with the same desire that we bring to launch. Industry has an answer to the AI power crisis. It has yet to produce an equally convincing answer to its sustainability problem. Organizations that succeed in orbit will be the ones that solve both.
