Cloud computing was built around an assumption that the infrastructure supporting digital services will remain on Earth. That assumption is beginning to change.
China has now put into regular operation what Beijing University of Posts and Telecommunications (BUPT) describes as the world’s first orbital computing. cloud. The platform connects satellite computing resources, ground stations and terrestrial data centres, allowing users to submit tasks, deploy software, monitor execution and receive results. BUPT says it has already completed hundreds of orbital computing calls for more than 100 users.
The significance is not simply that computers are being placed on satellites. Satellites have carried processors for decades. What is different is the attempt to turn computing in orbit into a service.
Traditional satellite systems generally collect data and send it to Earth for processing. Orbital computing reverses part of that model. Data can be processed closer to where it is generated, reducing the need to transmit large volumes of information to terrestrial facilities and allowing decisions to be made more quickly. BUPT says its system is already being used for big-data processing, 6G communications trials and distributed-storage experiments. This could become more important as artificial intelligence expands the demand for computing power.
Google is investigating a similar direction through Project Suncatcher, which explores solar-powered satellite constellations carrying AI processors and connected by optical links. The attraction is clear: computation could be distributed across orbital infrastructure rather than being concentrated entirely in terrestrial data centres.
The argument for moving computing into space is also familiar. Land, electricity, cooling and water are becoming constraints on the expansion of large AI data centres. Space offers abundant solar energy and avoids some of the land-use and local resource pressures associated with terrestrial facilities. It also provides a global vantage point for processing data generated by satellites and other space-based systems.
The future cloud may not simply be distributed geographically. It may be divided geopolitically.
Cloud computing was built around an assumption that the infrastructure supporting digital services will remain on Earth. That assumption is beginning to change.
China has now put into regular operation what Beijing University of Posts and Telecommunications (BUPT) describes as the world’s first orbital computing(Opens in new window) cloud. The platform connects satellite computing resources, ground stations and terrestrial data centres, allowing users to submit tasks, deploy software, monitor execution and receive results. BUPT says it has already completed hundreds of orbital computing calls for more than 100 users.
The significance is not simply that computers are being placed on satellites. Satellites have carried processors for decades. What is different is the attempt to turn computing in orbit into a service.
Traditional satellite systems generally collect data and send it to Earth for processing. Orbital computing reverses part of that model. Data can be processed closer to where it is generated, reducing the need to transmit large volumes of information to terrestrial facilities and allowing decisions to be made more quickly. BUPT says its system is already being used for big-data processing, 6G communications trials and distributed-storage experiments. This could become more important as artificial intelligence expands the demand for computing power.
Google is investigating a similar(Opens in new window) direction through Project Suncatcher, which explores solar-powered satellite constellations carrying AI processors and connected by optical links. The attraction is clear: computation could be distributed across orbital infrastructure rather than being concentrated entirely in terrestrial data centres.
The future cloud may not simply be distributed geographically. It may be divided geopolitically.
The argument for moving computing into space is also familiar. Land, electricity, cooling and water are becoming constraints on the expansion of large AI data centres. Space offers abundant solar energy and avoids some of the land-use and local resource pressures associated with terrestrial facilities. It also provides a global vantage point for processing data generated by satellites and other space-based systems.
But moving the cloud into orbit does not make it less vulnerable to competition between countries. It may make it more so.
On 3 September, four US space companie named as SpaceX, Blue Origin, Stoke Space and Starcloud withdrew from French President Emmanuel Macron’s forthcoming space summit after, according to French officials and reporting by POLITICO, pressure from the Trump administration. Starcloud is particularly relevant because it is developing space-based data-centre technology.
The immediate dispute concerns European strategic autonomy. France and other European governments have been seeking greater independence in launch, satellite communications and other parts of the space economy, while Washington has pushed back against European measures that could disadvantage US companies. That dispute points to a larger problem for orbital computing. The future cloud may not simply be distributed geographically. It may be divided geopolitically.
Today, a company can build a data centre in one country, operate cloud infrastructure across several jurisdictions and connect users through a global communications network. Orbital computing adds another jurisdictionally complicated layer. A workload could be generated by a satellite operated in one country, processed by computing hardware in orbit owned by a company from another, routed through ground stations elsewhere and ultimately delivered to a terrestrial data centre elsewhere.
This leaves questions about the jurisdiction over data being processed by an orbital computer, which country’s laws apply to a satellite providing cloud services over multiple regions, or who is responsible when a service fails.
Space is already a contested environment. Satellites support communications, navigation, intelligence, surveillance and military operations. States have developed or demonstrated capabilities to interfere with space systems through jamming, cyber operations, dazzling and other forms of disruption. Orbital computing would add another layer of valuable infrastructure to that environment. Its vulnerability would not necessarily come only from an attack on the computing hardware itself.
An orbital cloud revolves around the chain of systems – satellites, software, optical and radio communications, ground stations, terrestrial data centres, power systems and launch infrastructure – and if any part of it is disrupted, whether via physical harm to a satellite or a cyberattack against an orbiting computing platform, the impact cascades. This is where orbital computing deviates from the conventional data centre. An attack on a terrestrial facility is usually limited to an area, but damage in orbit can spread debris, threatening other satellites –including ones owned by states and businesses not involved in the initial disagreement.
The answer should not be to stop the advancement of orbital computing. It has a lot of potential advantages. Reducing communications bottlenecks, facilitating quicker decision-making, and enabling more complex space-based applications are among potential benefits of processing satellite data in orbit.
Governments should not make the mistake of viewing cloud infrastructure in space as solely commercial. Operators should be obliged to reveal basic information about satellite numbers, orbital locations, ownership, manoeuvrability, communications dependencies, and deorbit plans prior to the widespread use of orbital computing networks. Orbital cloud providers should be subject to the same cyber resilience and incident reporting standards as are increasingly being applied to vital terrestrial infrastructure.
More clarity is also required regarding wartime legal status. International law, where possible, ought to differentiate between systems directly supporting military activities and civilian orbital computer infrastructure. In situations when the loss of one computer platform could put unrelated spacecraft in peril, operators should have explicit responsibilities regarding debris mitigation and collision avoidance. Communication during a crisis will be just as crucial. Mechanisms for quickly identifying if an unanticipated orbital occurrence is accidental, technical, or intentional are needed by governments, satellite operators, cloud firms, and space organisations.
Whether orbital computing is technically or financially feasible is no longer the only concern. It is whether the world is prepared for a future in which cloud infrastructure becomes part of contested space infrastructure.
