Space is no longer just a place we look at. It’s becoming a place we compute in.
With SpaceX’s launch capabilities making low-Earth orbit commercially viable at scale, the conversation about placing compute infrastructure in space with orbital data centres has shifted from hypothetical to strategic. Whether space-based data centres will exist is no longer the question worth asking. The more important question is whether the infrastructure that enables them will serve everyone equitably, or consolidate power in the hands of those who already own the rocket.
That question requires honesty about both the genuine technical case for space compute and the structural risks embedded within it.
Why space actually makes sense for compute
The case starts with the physics. Low-Earth orbit offers something terrestrial data centres have spent decades and billions trying to engineer: natural thermal management. The extreme temperature differential between sunlight and shadow, combined with the vacuum of space, creates conditions that could dramatically reduce the cooling overhead that typically consumes roughly 40% of a data centre’s energy budget. That is a fundamental rethinking of one of the industry’s most persistent cost problems.
Then there is the question of power. Solar energy in orbit is uninterrupted, unfiltered and abundant in ways that ground-based solar cannot match. A space-based data centre could theoretically access eight times the solar energy available at Earth’s surface, without seasonal variation, cloud cover or land-use constraints. For an industry under mounting pressure to decarbonise, that is a meaningful proposition.
And finally, there is latency. As edge computing matures, the ability to process data at or near the point of collection becomes critical. For satellite-generated data covering environmental monitoring, precision agriculture, defence and intelligence applications, and logistics, processing in orbit rather than routing back to a ground station could deliver latency advantages that reshape what is possible in real time. The commercial applications are rapidly catching up with what specialist sectors have understood for years.
Taken together, these are technically sound arguments, and the case for space-based compute is anything but speculative marketing.
The sustainability question: where it holds up, and where it doesn’t
The sustainability argument is more nuanced than the headlines suggest. On the one hand, eliminating conventional cooling energy consumption, accessing near-infinite solar power and reducing the land footprint of digital infrastructure are all meaningful environmental gains. For an industry that already consumes a significant and growing share of global electricity, with AI workloads accelerating that trajectory, any structural reduction in energy intensity matters.
On the other hand, getting infrastructure into orbit requires rockets, and rockets burn fuel. Even with reusable launch vehicles bringing costs down considerably, the carbon cost of a single launch remains substantial. If space data centres require frequent hardware refreshes – and the current pace of AI hardware development suggests they will – the cumulative launch emissions could quickly offset the operational gains. The growing problem of orbital congestion and space debris adds further long-term environmental and operational risk that the industry has yet to fully reckon with.
Space compute could be sustainable at scale, but only if launch frequency is managed carefully, hardware longevity is prioritised and full lifecycle emissions are properly accounted for. Currently, none of those conditions are guaranteed. The sustainability case is real, but it requires scrutiny rather than assumption.
The structural question the industry needs to address
Some of the organisations best positioned to develop orbital compute are those that already own significant portions of the surrounding infrastructure stack, including launch capability, low-Earth orbit connectivity, and now potentially the processing layer that sits between data and insight. That kind of vertical integration is commercially logical. From an infrastructure governance perspective, it also raises legitimate questions.
When the same provider owns launch, communication and computation, it owns the stack, and whoever owns the stack sets the terms, including pricing, access conditions, geographic availability, data sovereignty standards and who gets to build on top of it.
For AI specifically, this matters more than most coverage acknowledges. The future of AI is not just about who has the best models, but about who has access to the compute and data pipelines to train and run those models at scale. If the infrastructure layer enabling space-based AI compute is controlled by a small number of vertically integrated actors, the result is not a neutral utility. It is a toll road with very few entry points.
We have been here before with terrestrial cloud when the concentration of cloud infrastructure in the hands of a small number of providers has concerned regulators, governments and enterprises for years. The Competition and Markets Authority, the European Commission and sovereign cloud initiatives like EURO-3C are all responses to the same underlying problem: critical digital infrastructure in private hands, without meaningful competition, creates dependencies that are very hard to unwind.
Space compute risks concentrating that problem further, and unlike terrestrial alternatives, there is no on-premises equivalent for orbit.
The infrastructure imperative
None of this is a case against space-based data centres, rather it is a case for thinking carefully about how they should be built, governed and managed.
Digital infrastructure decisions made early tend to stick, and the contrast is instructive: the internet’s open architecture was a deliberate choice, just as closed platform models were equally deliberate. The decisions being made now about who builds, owns and operates orbital compute infrastructure will shape what is possible for decades.
For organisations that depend on digital infrastructure, the lesson from the last decade of cloud adoption is straightforward: concentration in any single provider creates compounding risk. That applies with even greater force when a provider also controls the transport layer and the launch capability.
Addressing this requires open standards and interoperability built in from the start. It requires regulatory frameworks that treat orbital compute as critical international infrastructure rather than a purely commercial asset. It requires infrastructure professionals with the expertise to manage hybrid estates spanning terrestrial and orbital environments, without being locked into a single vendor’s roadmap. And it requires a genuine commitment to the proposition that managing digital infrastructure well, wherever it sits, is what determines whether a technology delivers broad benefit or narrow advantage.
The decisions made now will define who benefits later
The space data opportunity is real, the engineering case for putting compute closer to where data is generated is sound, and the potential gains in energy efficiency, latency and analytical depth are substantive.
But the ownership and governance question carries equal weight to the engineering one. If space-based data centres are built as extensions of tightly integrated proprietary platforms, the outcome will be extraordinary capability available on those platforms’ terms. If they are built on open, interoperable and independently managed foundations, the outcome looks considerably different.
The organisations and policymakers who engage seriously with that distinction now, while the architecture is still being designed, are the ones best placed to influence how it resolves. Waiting for the infrastructure to be built and then asking who it serves is a concession the industry cannot afford to make.
Chris Carreiro
Chris Carreiro is CTO, at Park Place Technologies where he oversees corporate innovation, research and development, and the creation of new portfolio offerings.
As one of the founding figures behind ParkView, Chris played a central role in Park Place’s evolution into a data centre services company. He brings deep expertise in emerging technologies, including edge computing, AI, blockchain and liquid immersion cooling.
