The Space Reviewin association with SpaceNews
 


 
SpaceX AI sat
SpaceX has proposed launching as many as one million orbital data center satellites, and other companies have proposed constellations of thousands of such satellites. (credit: SpaceX)

Space race or space divide: orbital AI and the Global South’s exclusion crisis


Several months ago, Elon Musk announced that SpaceX, recently merged with his artificial intelligence company xAI, would place data centers into orbit around the Earth. This announcement was not an isolated development. Google had already unveiled its Project Suncatcher in November 2025, with plans to launch approximately 80 data-center satellites into orbit by 2027. The startup Starcloud successfully trained the first AI model in space in December 2025.

The argument goes that moving computation to orbit alleviates pressure on terrestrial infrastructure. But this framing conveniently ignores who controls the infrastructure being proposed as the solution, and at whose expense.

What began as a speculative idea has rapidly become a corporate race. This race, however, is not being run on a neutral track. Low Earth orbit (LEO) is a finite and already severely saturated resource, mostly by the Global North. The nations of Global South, which have historically had the least access to it, stand to lose the most as wealthy Western corporations plan to further capture what’s left of the accessible space. The decision to put AI data centers in space can have profound consequences for global equity, international security, and the governance of a resource that in principle belongs to all of humanity.

The appeal of space-based AI infrastructure is straightforward on its face. On Earth, the infrastructure demands of large-scale AI are straining power grids, depleting water reserves used for cooling, and consuming land. A single AI-focused hyperscale data center can consume as much power as 100,000 homes. All this is happening at a pace that has triggered regulatory and public resistance, particularly in the United States.

The United States is not alone in this calculus. China has pursued parallel orbital computing initiatives as part of its broader space strategy. Proponents argue that orbital facilities offer perpetual solar energy, reduced latency for certain satellite-based applications, and a way to sidestep the terrestrial constraints that are slowing AI development on the ground. The argument goes that moving computation to orbit alleviates pressure on terrestrial infrastructure. But this framing conveniently ignores who controls the infrastructure being proposed as the solution, and at whose expense.

Within the United States, there are voices urging a pause. Recently, in March 2026, Sen. Bernie Sanders and Rep. Alexandria Ocasio-Cortez introduced the Artificial Intelligence Data Centre Moratorium Act, legislation that would halt the construction of new AI data centers until “strong national safeguards” are in place. The bill gestures at a deeper contradiction that its authors may not have fully intended to surface. The primary driver behind moving data centers to orbit is that terrestrial resources in the United States are becoming too constrained to sustain the pace of AI expansion. Yet the orbital environment into which these corporations now propose to expand is not American territory.

Earth orbit, under Article II of the Outer Space Treaty 1967, cannot be claimed by any nation through use or occupation. The treaty designates outer space, including low Earth orbit, as the province of all mankind. If the reason for this expansion is a lack of domestic resources, then the proposed solution is to use a resource shared by all nations without most countries’ consent. The Global South has no seat at the table where these decisions are being made. No multilateral body has been convened to ask whether humanity collectively endorses this use of its shared orbital commons or not. The capital belongs to the handful of corporations involved, and the regulatory environment is similarly being dominated by a few. The resource being consumed, however, belongs to everyone.

No multilateral body has been convened to ask whether humanity collectively endorses this use of its shared orbital commons or not. The capital belongs to the handful of corporations involved, and the regulatory environment is similarly being dominated by a few.

The implications for the Global South are particularly acute. There is a version of the orbital data center argument that presents itself as an opportunity for the Global South. If computation moves to orbit, developing nations could simply purchase access to that infrastructure rather than building their own. This argument deserves to be taken seriously, and then frankly rejected. The Iran-United States conflict has offered a clarifying lesson. Commercial satellite operators, under pressure from the United States government, halted imagery sales of the conflict region. The episode demonstrated that access to privately operated space infrastructure is not a market transaction insulated from geopolitics. It is a relationship of dependency that can be severed at any moment by the political decisions of a more powerful state.

Countries across the Global South that rely on orbital data centers controlled by American, Chinese, or any other powerful nation’s corporations would face the same vulnerability in any future conflict. This is precisely why demand for sovereign satellite capabilities may intensify among developing nations in upcoming years. An orbital data center ecosystem dominated by a handful of powerful nations does not extend sovereignty to others. Instead, it deepens the dependency that already defines the relationship between the Global North and the Global South in technology, finance, and now, space. If the most accessible orbital altitudes become saturated with AI computing infrastructure controlled by a handful of corporations, the sky above the countries of the Global South will effectively be privatized by foreigners.

LEO is also not an infinite commons. It is a finite and already severely congested resource. The Kessler Syndrome is no longer a theoretical physicist’s concern. As of 2025, the number of tracked objects in low Earth orbit had risen to over 24,000, an increase of 76% since 2019. SpaceX’s Starlink satellites alone performed 145,000 collision-avoidance maneuvers in just six months prior to July 2025. The European Space Agency has warned that in certain altitude bands, the density of active objects is now comparable to that of space debris. Against this backdrop, SpaceX has announced plans to launch one million satellites to form an orbital data center network. Adding a vast layer of computing infrastructure atop an already overcrowded orbital environment can further reduce access for the Global South.

Orbital congestion has material consequences, especially for the Global South. Nations seeking to deploy satellites for remote sensing, weather monitoring, agricultural planning, or telecommunications will encounter an environment made more hazardous and more costly by the prior decisions of those in the private offices of Global North. Global South countries, many of which lack the independent launch capability and depend on foreign launch providers, would be disproportionately harmed.

Will there be enough space in LEO when the Global South becomes capable of launching its own satellites or data centers? The answer seems grim as even active objects in such orbits are being compared to space debris. This means that ongoing space race, which now includes orbital data centers, may take away the Global South’s claim in a common resource that is outer space.

The race to put AI data centers in orbit is being framed as an inevitable march of progress. But progress toward what, and for whom?

What is needed is a fundamental rethinking of the governance framework for outer space AI infrastructure before the changes being proposed by companies become irreversible. The designation of space as humanity’s common resource has never been operationalized in a way that gives developing nations a meaningful say over how orbital resources are allocated or protected. A moratorium on large-scale orbital data infrastructure, pending the development of an inclusive international regulatory framework, is not an unreasonable demand. It is the minimum standard of due diligence that a genuinely global technology should require, if it truly is global in nature.

The race to put AI data centers in orbit is being framed as an inevitable march of progress. But progress toward what, and for whom? The nations of the Global South did not create the AI systems that require this infrastructure and they certainly had no chance to accumulate the capital that is funding it. These countries will not own the data centers when they are in orbit. And they will not have a meaningful voice if those centers become targets in a conflict between great powers.

The sky above us all is being claimed, incrementally and without consent, by a handful of corporations accountable to their shareholders and to no one else. That is not the future of humanity. It is the future of a few, imposed on the rest. The necessary dialogue and legislation to end this structural discrimination has to be done before the sky above the Global South is claimed by a few private companies.


Note: we are now moderating comments. There will be a delay in posting comments and no guarantee that all submitted comments will be posted.

Home