The trickier issue is governance – how these assets will be treated once they become part of a contested space environment.
Drone attacks on data centre facilities in the United Arab Emirates and Bahrain during US–Israel–Iran conflict underscore that digital infrastructure can be physically exposed in war. The cloud is not abstract. It has buildings, power lines, cooling systems, cables, staff, and physical vulnerabilities. When these systems support communications, logistics, government services, or military activity, their strategic relevance increases.
Vulnerabilities of this type have led to orbital data centres being floated as an attractive option. This goes beyond physical security concerns to the impact resource-intensive data centres can have on communities in the AI age. Solar power is more accessible in space, while placement in orbit escapes land-use disputes and direct pressure on local water systems. The technological hurdles are real. Radiation can damage chips. In space, heat is hard to dissipate. Repairs are costly. Optical links must remain stable. Launch and replacement expenses remain high. If companies can overcome these hurdles, putting data centres in orbit might ease certain stresses on Earth.
Google’s Project Suncatcher is investigating solar-powered satellite constellations with TPU AI processors and free-space optical links as a means to scale machine-learning computation in space. SpaceX has shown interest in building artificial intelligence data centres in space, even though the company has reportedly cautioned investors against backing these projects given the untested technology. Meta, while not formally proposing an orbital data centre, has teamed up with Overview Energy to investigate space-based solar power for data centres, with a demonstration in orbit scheduled for 2028.
But while this possible movement into orbit is often discussed in terms of energy access, scalability, technical reliability and as an answer to growing community resentment, it does not solve the question of how space-based data centres would be treated during conflict. If businesses begin transferring data-centre operations into orbit, the security risks would be considerable.
An attack on an orbital data centre would not only create economic disruption but could also increase the risk of broader escalation.
Space is already a contested environment, where satellites aid surveillance, communication, navigation, targeting, and battle damage assessment. In times of crisis, adversaries may attempt to jam, spoof, hack, dazzle, or disable them because of the military advantage they provide. This contested environment would also house orbital data centres. A ground attack on a data centre can destroy servers and interrupt services. An attack, collision or debris event in orbit can threaten unrelated satellites for years. Weather forecasting, communications, navigation, Earth observation, disaster response, and commercial internet services could be affected even if they had no role in the original conflict.
The space economy now consists of an upstream sector, which includes satellite manufacturing, launches, and ground infrastructure, and a much larger downstream sector, which includes GPS, communications, Earth observation, weather services, and other satellite-enabled applications. According to one estimate, direct turnover from space-related activities has already been recorded at around $470 billion to $670 billion. This is expected to reach around $1 trillion, and some projections place the wider space economy even higher by 2035.
An attack on an orbital data centre would not only create economic disruption but could also increase the risk of broader escalation if states interpret the attack as part of a wider campaign against their space infrastructure.
The debate should move beyond whether orbital data centres can be commercially certified as viable. The trickier issue is governance should they become part of a contested space environment.
Satellite counts, orbital shells, manoeuvrability, deorbit timelines, spectrum use, collision-avoidance systems, and ground-station dependencies should all be made public by operators. Four areas then need priority.
First, wartime legal status should be clarified before deployment, not after a crisis. A purely civilian orbital data centre should not be treated the same as infrastructure directly supporting battlefield targeting and command systems.
Second, debris accountability should be stronger because damage in orbit can spread beyond the original target and threaten third-party satellites for years.
Third, cyber resilience and service continuity should be mandatory, since orbital compute would be a high-value target for disruption.
Fourth, crisis communication channels should connect states, cloud providers, and satellite operators so that a malfunction, cyber incident, or debris event is not mistaken for a deliberate attack.
Artificial intelligence, climate modelling, research, and resilient storage might all benefit from data centres in orbit. Yet the recent disruption of cloud infrastructure on Earth shows why the security question cannot wait. Although relocating data centres to space would alleviate some political pressure on Earth, it would also put AI infrastructure in a domain already shaped by military competition.
