Why SpaceX’s bid to orbit a million data centres is not science fiction — and why it should concern every IT decision-maker on Earth
On 30 January 2026, SpaceX filed a landmark application with the Federal Communications Commission requesting authority to launch up to one million satellites. But this time it’s not for broadband connectivity, but to operate what the company calls the SpaceX Orbital Data Center system. The filing described a solar-powered constellation operating between 500 and 2,000 kilometres in altitude, relying almost exclusively on high-bandwidth optical inter-satellite links. In the company’s own words, the initiative is “a first step towards becoming a Kardashev II-level civilisation — one that can harness the Sun’s full power — while supporting AI-driven applications for billions of people today.”[1]
The ambition is breathtaking. There are currently approximately 15,000 artificial satellites orbiting Earth; SpaceX’s proposal would increase that number by a factor of seventy.[2] SpaceX is not alone: Blue Origin has announced its TeraWave constellation of some 5,400 networking satellites, China has revealed a 200,000-satellite constellation focused on in-orbit processing for sensitive state applications, and Starcloud submitted its own FCC proposal for up to 88,000 orbital data centre satellites in February 2026.[3] Computing is heading skyward, and the drivers are structural, economical, and political.
The Pressure on Earth’s Infrastructure
Artificial intelligence is consuming energy at a pace that terrestrial grids cannot sustain. The International Energy Agency projects that global data center electricity consumption will surpass 1,000 terawatt-hours by the end of 2026.[4] That demand runs headlong into constrained land, limited water resources for cooling, and electricity grids already under strain. Space sidesteps all three constraints simultaneously. Orbital altitudes offer uninterrupted solar exposure: no day-night cycle interruption, no seasonal variation, no grid dependency. Waste heat is rejected by radiating directly into a vacuum at approximately 2.7 Kelvin. There is no permitting, no neighbourhood opposition, no water license to renew. At least at a first glance.
SpaceX made the economic argument explicit in its FCC filing: “The capacity for intelligence processing in space could surpass the electricity consumption of the entire U.S. economy, without the immense cost and disruption of rebuilding Earth’s strained electrical infrastructure.” If Starship can reduce launch costs to the company’s target of ten to twenty dollars per kilogram — against a current Falcon 9 rate of roughly 2,700 dollars per kilogram — the economics begin to look genuinely competitive for AI training workloads.
The Technical and Regulatory Friction
Translating the vision into operational infrastructure is another matter. In the vacuum of space, convective and evaporative cooling are impossible; heat can only be rejected by thermal radiation. For a 1-megawatt orbital data center, back-of-the-envelope physics demands approximately 1,600 square meters of radiator surface — roughly the footprint of a hockey rink.[5] Cosmic radiation in low Earth orbit generates single-event upsets, bit flips, and cumulative transistor degradation. Traditional radiation-hardened silicon costs up to 600 times more than commercial-grade chips and lags a decade behind in processing performance. The connectivity bottleneck is equally real: even at LEO altitudes, round-trip latency adds overhead that rules out time-sensitive applications, limiting the orbital data center to batch workloads such as model training, scientific simulation, and genomic processing.
The regulatory environment presents its own contradictions. The FCC, unlike the FAA or NASA, has historically not conducted systematic environmental reviews for satellite constellations — a gap that a 2022 U.S. Government Accountability Office audit explicitly flagged. Debris accumulation, stratospheric alumina from rocket reentries, and light pollution from reflective surfaces are all peer-reviewed concerns that regulators have yet to address at scale. The FCC recently authorised SpaceX to launch an additional 7,500 Starlink satellites while deferring judgement on a further 14,988; a sign that even incremental progress faces institutional friction.
The Legal Vacuum as a Business Opportunity
Yet the most consequential dimension is jurisdictional. On Earth, data sovereignty is geographic: a server in Frankfurt obeys German law; a server in Virginia obeys U.S. law. In orbit, a satellite travels at 7.6 kilometres per second, overflying Brazil, the Atlantic, and France in the time it takes to read this paragraph.[6] The 1967 Outer Space Treaty, which is still the foundation of space law nowadays, designates outer space as the “province of all mankind” and prohibits territorial sovereignty claims, but says nothing about commercial data processing.
Under current doctrine, the most workable framework is the “digital flag state” model: a satellite is treated as an extension of its registration state, the same way a ship is treated under maritime law.[6] A U.S.-registered orbital data centre falls under U.S. federal law; one registered in Luxembourg obeys Luxembourg’s space asset legislation. This creates a commercially significant opportunity for jurisdictions with favourable privacy, tax, or national-security regulations to attract orbital infrastructure: not unlike the financial services dynamic that drove offshore registrations for decades. For enterprises handling sensitive health, financial, or defence data, an orbital platform registered in a permissive jurisdiction could, in principle, operate outside the reach of GDPR, CLOUD Act demands, or local law enforcement access. This represents a capability with obvious appeal and obvious ethical complexity.
The European Space Agency’s ASCEND programme (Advanced Space Cloud for European Net zero emissions and Data sovereignty), funded at 300 million euros through 2027, is already positioning the EU as a counterweight, with a demonstration mission planned for 2026.[7] The EU Space Act, adopted in June 2025, further signals that Europe intends to establish its own governance framework before the orbital infrastructure landscape consolidates around U.S. and Chinese actors.
The conversation about data residency, compliance architecture, and infrastructure strategy is about to gain a vertical dimension.
References
1. SpaceX FCC Filing — Fortune, February 2026
2. Astrobites — Two Satellite Proposals Threaten Dark and Quiet Skies, February 2026
3. Wikipedia — Space-Based Data Center
4. The Next Web — SpaceX S-1 Warns Orbital AI Data Centres May Not Be Viable, May 2026
5. World Economic Forum — Why Cooling Is the Real Obstacle to Space-Based Data Centres, 2026
6. Bloomberg Law — Digital Flag State Rule Would Give Space Law a Regulatory Boost, January 2025
7. AI News Hub — Space-Based Data Centres: The Future of AI Computing, 2025
