Orbital Data Centres and the Limits of National Jurisdiction over Technology Firms

The largest technology firms are moving their data centres into low Earth orbit, having pushed terrestrial infrastructure to the physical and environmental limits of what it can supply in power, cooling and land to an artificial intelligence sector whose demands keep compounding. SpaceX has set the fourth quarter of 2027 for the launch of its first generation of computing satellites, targeting one gigawatt of orbital data centre power capacity in that same year and one hundred gigawatts by 2030. The scale of the pressure behind that schedule is visible in the wider market, where global data centre demand is estimated to require investment approaching €5.7 trillion before the decade closes.

What presents itself as an engineering migration is in substance a redistribution of sovereign authority. Moving processing beyond national territory removes technology firms from the reach of the data localisation rules, compliance regimes and tax frameworks that states have spent a decade constructing. For Arab states the challenge is immediate. Having invested heavily in domestic digital infrastructure and imposed strict data residency requirements, they now face the prospect that the server handling their citizens’ data sits five hundred kilometres overhead, beyond the enforcement reach of their courts.

This analysis therefore examines the economics of moving computation into orbit and the limits of its viability; traces the legal gap that allows technology firms to shelter behind the jurisdiction of the state of registry in order to avoid the laws of every other state; assesses the risks of monopolistic concentration and the environmental costs borne collectively; and identifies the regulatory instruments available to Arab states in defence of their digital sovereignty, chief among them the management of radio-frequency spectrum, satellite landing rights and the supervision of ground gateways.

The physical constraints of terrestrial infrastructure are pushing the computing industry to look for alternatives beyond the atmosphere. Conventional data centres contend with land scarcity, unstable local electricity grids and the vast volumes of water required to carry heat away, at precisely the moment when computing workloads are multiplying. An artificial intelligence agent generates roughly 450 per cent more network traffic than a human user performing the same task, and that multiplier compounds with every layer of automation added to enterprise systems. Space offers the inverse profile: low ambient temperature, uninterrupted solar exposure and no residential constituency to object to resource consumption.

SpaceX leads the field, having accelerated its own timetable more than once in a single year. The company is targeting large-scale deployment of these servers by 2028, working towards a constellation of up to one million satellites, each functioning as a data centre in its own right. Its hardware rests on an exclusive partnership with Nvidia to adapt rack-scale computing systems for orbital operation, built around central processing units carrying 88 cores with memory bandwidth of 1.2 terabytes per second, and graphics processing units that the company rates at twenty-five times the performance of the preceding terrestrial generation.

The spacecraft itself is a single structure twenty metres in height, with a solar array span of seventy metres, a peak generating capacity of 150 kilowatts and an average computing load of 120 kilowatts. Reaching one gigawatt on that basis requires the manufacture of some 7,400 server racks a year, a figure that has already prompted major contract manufacturers such as Foxconn and Quanta to prepare for the demand. The industrial bottleneck, in other words, has shifted from the launch pad to the assembly line.

Engineering requirements in orbit differ fundamentally from those on the ground. In sun-synchronous orbits between 500 and 2,000 kilometres, solar irradiance holds steady at 1,361 watts per square metre, uninterrupted by night or cloud. The absence of air, however, rules out convective cooling, leaving radiative cooling governed by the Stefan–Boltzmann law as the only mechanism available: a one-megawatt facility needs roughly 1,600 square metres of thermal radiators, though it consumes no water at all. Economic viability consequently rests on two conditions. Launch costs must fall to between $100 and $500 per kilogramme aboard Starship, and the premium for radiation hardening must continue its decline from a former range of five to ten times conventional silicon to about one and a half times, a shift made credible by proton-beam testing of shielded commercial chips.

That gap in capital expenditure explains why the field is confined to one firm, or at most two. Building a gigawatt of orbital capacity is put at some $170 billion against roughly $14.1 billion for its terrestrial equivalent, a ratio no ordinary market entrant can absorb. Those who defend the economics argue that the decisive advantage lies not in cost but in speed of deployment: connecting a terrestrial data centre to the electricity grid takes between five and seven years, whereas orbital capacity can be placed in service within months. In the market for artificial intelligence, that interval is worth entire generations of models.

Global data protection rests on a single premise: that legal jurisdiction over data follows the geographical location of the server that stores or processes it. The European Union’s General Data Protection Regulation, the United Arab Emirates’ Federal Decree-Law No. 45 of 2021 and Saudi Arabia’s Personal Data Protection Law all proceed from that assumption. Moving data centres into orbit severs the processing site from national territory altogether, because Article II of the 1967 Outer Space Treaty prohibits national appropriation of outer space by claim of sovereignty, by use or by occupation. A requirement that data remain within national borders is therefore unenforceable, in engineering terms as much as in legal ones, five hundred kilometres above them.

The gap makes room for an arrangement that closely resembles flags of convenience in merchant shipping. Article VIII of the Treaty, reinforced by the 1975 Registration Convention, vests jurisdiction and control over a space object solely in its state of registry, so that a satellite becomes an extension of that state’s sovereignty wherever it passes over the Earth. In practice, sensitive financial or health data sent by a firm in the Middle East or Africa to a server registered in the United States falls under American law alone, and the company hosting it is released from local compliance requirements and from restrictions on cross-border data transfers. Article VI offers no remedy: it obliges the state of registry to exercise continuing supervision over its non-governmental entities, but grants the state of the data subject no right of recourse whatsoever.

The severity of the gap is compounded by the collapse of enforcement tools. For two decades, the capacity of states to discipline the digital economy has rested on physical access: inspecting servers, seizing drives, throttling fibre-optic connections and pursuing local executives. None of these instruments reaches a server travelling at 27,000 kilometres an hour. Inaccessibility is moreover being converted into a deliberate commercial feature. New security architectures generate encryption keys inside secure elements after launch, so that neither manufacturer nor operator holds them and neither can be compelled by a court to surrender them. At end of life the facility burns up entirely during atmospheric reentry, erasing the data permanently and leaving nothing that any authority could examine or audit.

The consequences extend to public revenue. International tax treaties and the Pillar One initiative within the OECD’s Base Erosion and Profit Shifting project both turn on the concept of the permanent establishment, meaning a fixed place through which business is carried on, such as an office, a factory or a server hall. That basis disappears when a model is trained on data drawn from Arab, African or Asian markets while the computation itself takes place on an object that crosses international borders every few minutes. Even digital services taxes, levied unilaterally by reference to the location of the user rather than the server, are difficult to enforce where the taxing state can identify no assets, bank accounts or ground gateways to attach. The European Commission alone has legislated in response, through the draft Space Act of 2025, whose Article 27 requires that space data offered within the Union originate from objects entered in its own registry, a regional shield that leaves emerging markets uncovered.

The shift carries risks of economic concentration beyond anything the digital economy has produced so far. Terrestrial cloud operators are enormous, yet they still depend on independent firms for electricity, land development and fibre-optic transit. In orbit, a single operator can own every stage of service delivery through complete vertical integration: the only launch vehicle capable of lifting the payload at the necessary price, orbital slots reserved in advance through licence applications running to a million satellites, the inter-satellite laser network operating at a terabit per second, and the computing hardware itself. Leading artificial intelligence laboratories have already begun reserving portions of that capacity under lease agreements worth billions of dollars, which hands the operator considerable power over the price of computation and over the pace at which frontier models can advance anywhere in the world.

The costs, by contrast, fall on the global commons, meaning those shared domains and resources that lie outside the sovereignty of any single state, outer space among them. Deploying tens of thousands of large satellites raises the probability of a cascading sequence of collisions known as the Kessler Syndrome, which could fill critical orbits with lethal debris and disable communications and Earth observation for generations. Atmospheric effects follow directly from the same deployment. Because atmospheric drag limits the working life of these spacecraft to between three and five years, they must be replaced continuously, and the retired units burn up in the upper atmosphere, leaving aluminium oxide nanoparticles that catalyse the chlorine reactions responsible for ozone destruction. The Federal Communications Commission nonetheless continues to operate under a categorical exclusion that allows it to license satellites without conducting any environmental impact assessment.

These developments push Arab states towards regulating the path data travels rather than the site at which it is processed. The dependence of orbital data centres on a link to the ground is their single point of vulnerability, since processed data has no value until it reaches a user or a client. That link necessarily relies on radio-frequency spectrum governed by the Radio Regulations of the International Telecommunication Union, which prohibit the establishment or operation of any transmitting station without a licence from the state within whose jurisdiction the station lies.

The connection to the ground gives the Communications, Space and Technology Commission in Saudi Arabia and the Telecommunications and Digital Government Regulatory Authority in the United Arab Emirates four practical instruments. The first ties satellite landing rights and spectrum allocation to an explicit contractual undertaking by the orbital operator to observe local data protection law and to shield citizens’ data from foreign access requests; refusing the licence amounts, in practice, to barring an entire satellite constellation from serving the Arabian Peninsula. The second requires operators to route commercial orbital traffic through sovereign ground gateways subject to state audit rather than beaming it directly to user terminals, which restores a physical control point at which firewalls, cross-border transfer assessments and tax collection can be applied. The third conditions market access on partnership with a national telecommunications entity that retains the commercial relationship and the decryption keys, so that the orbital operator functions as a wholesale supplier rather than a direct monopolist over the consumer. The fourth turns outward, towards an Arab negotiating bloc within the International Telecommunication Union and the Committee on the Peaceful Uses of Outer Space, pressing for orbital data centres to be classified as a legal category distinct from communications satellites, for the immunity of the state of registry to be curtailed where the data processed originates elsewhere, and for disputes to be referred to the 2011 PCA Optional Rules for Arbitration of Disputes Relating to Outer Space Activities rather than to the courts of the operator’s own state.

Questions of digital sovereignty are therefore migrating from a contest over the location of the server to the regulation of the spectrum that connects it to the Earth, and the window for acting is narrowing. Capital costs and manufacturing constraints suggest that orbital data centre capacity genuinely in service will not exceed five to eight gigawatts by 2030, less than a tenth of the hundred gigawatts announced. Scarcity will thus be the rule, and the monopoly operator will retain the power to set prices. The regulatory window available to the Gulf states runs between 24 and 36 months, to the end of 2028, before the first commercial constellation enters service; after that point, amending licence conditions becomes a negotiation over arrangements already in place rather than rule-setting before the activity begins. Whether these instruments succeed will depend on a single calculation: that losing a Gulf market whose public and private sectors together spend tens of billions a year on cloud computing costs the orbital operator more than complying with its laws.

Humans of Technology. “Servers without Borders and Orbital Data Center.” Accessed September 6, 2026. https://humansoftechnology.com/en/servers-without-borders-and-orbital-data-center/.

European Space Policy Institute. “Data Centres in Space.” ESPI, November 2025. Accessed September 6, 2026. https://www.espi.eu/wp-content/uploads/2025/11/SBDCs.pdf.

Light Reading. “With Nvidia on Board, SpaceX Orbital Data Centers to Hit ‘Significant Scale’ in 2028.” Accessed September 6, 2026. https://www.lightreading.com/satellite/with-nvidia-on-board-spacex-orbital-data-centers-to-hit-significant-scale-in-2028.

Quartz. “SpaceX Aims to Launch Nvidia-Powered AI Satellites by Late 2027.” Accessed September 6, 2026. https://qz.com/spacex-nvidia-ai-satellites-orbit-2027-082526.

Luminix. “Data Centers in Space: Feasibility and Economics.” Mid-2026. Accessed September 6, 2026. https://www.useluminix.com/reports/industry-analysis/data-centers-in-space.

Blockchain Council. “Starcloud: Data Centre Infrastructure in Space — Valuation, Funding, Tech Roadmap.” Accessed September 6, 2026. https://www.blockchain-council.org/news/starcloud-data-centre-in-space-valuation-funding-tech-roadmap/.

Oracle. “Data Sovereignty vs. Data Residency: Three Key Differences.” Accessed September 6, 2026. https://www.oracle.com/ae/security/saas-security/data-sovereignty/data-sovereignty-data-residency/.

United Nations Office for Outer Space Affairs. “Treaty on Principles Governing the Activities of States in the Exploration and Use of Outer Space, Including the Moon and Other Celestial Bodies.” Accessed September 6, 2026. https://www.unoosa.org/oosa/en/ourwork/spacelaw/treaties/outerspacetreaty.html.

Oxford Public International Law. “Spacecraft, Satellites, and Space Objects.” Max Planck Encyclopedia of Public International Law. Accessed September 6, 2026. https://opil.ouplaw.com/display/10.1093/law:epil/9780199231690/law-9780199231690-e1222.

Dataversity. “Legal Issues for Data Professionals: Data Centers in Space.” Accessed September 6, 2026. https://www.dataversity.net/articles/legal-issues-for-data-professionals-data-centers-in-space/.

SpaceComputer. “Space Fabric: A Satellite-Enhanced Trusted Execution Architecture.” arXiv, 2026. Accessed September 6, 2026. https://arxiv.org/html/2603.23745v1.

KPMG. “Taxation of the Digitalized Economy: Developments Summary.” 2026. Accessed September 6, 2026. https://kpmg.com/kpmg-us/content/dam/kpmg/taxnewsflash/pdf/2026/digitalized-economy-taxation-developments-summary.pdf.

International Review of Public Justice. “Base Erosion and Profit Shifting Challenges in the Digital Economy.” Accessed September 6, 2026. https://irpj.euclid.int/articles/base-erosion-profit-shifting-challenges-in-the-digital-economy-in-sub-saharan-africa-a-framework-for-fairer-tax-treaty-cooperation/.

International Bar Association. “In Orbit: The Legal Implications of Data Centres in Space.” Accessed September 6, 2026. https://www.ibanet.org/in-orbit-the-legal-implications-data-centres-space.

TechRepublic. “Anthropic, SpaceX Deal Boosts Claude Compute and Points to Orbital Supercomputing.” Accessed September 6, 2026. https://www.techrepublic.com/article/news-anthropic-spacex-claude-compute-colossus-1/.

American Institute of Aeronautics and Astronautics. “Simulating the Evolution of Lethal Nontrackable Population for Low Earth Orbit.” Journal of Spacecraft and Rockets. Accessed September 6, 2026. https://arc.aiaa.org/doi/10.2514/1.A36198.

Ferreira, José P., et al. “Potential Ozone Depletion From Satellite Demise During Atmospheric Reentry in the Era of Mega-Constellations.” Geophysical Research Letters, 2024. Accessed September 6, 2026. https://www.researchgate.net/publication/381325029_Potential_Ozone_Depletion_From_Satellite_Demise_During_Atmospheric_Reentry_in_the_Era_of_Mega-Constellations.

Earthjustice. “Petition for a Programmatic Environmental Impact Statement.” July 8, 2026. Accessed September 6, 2026. https://earthjustice.org/wp-content/uploads/2026/07/2026.07.08-petition-for-programmatic-eis.pdf.

Office for Utility Regulation and Competition (OfReg). “Responses to ICT 2024-2.” January 10, 2025. Accessed September 6, 2026. https://www.ofreg.ky/viewPDF/documents/2025-01-10-15-04-26-Responses-to-ICT-2024—2—10-January-2025-1-.pdf.

PwC Middle East. “Secure Cloud Adoption: Central Banks and Secure Cloud Adoption.” 2025. Accessed September 6, 2026. https://www.pwc.com/m1/en/publications/2025/docs/central-banks-and-secure-cloud-adoption.pdf.

Anali Pravnog Fakulteta u Beogradu. “A Critical Analysis of the PCA Outer Space Rules.” 2026. Accessed September 6, 2026. https://anali.rs/xml/202-/2026c/2026-1c/Anali_2026-1c-03.pdf.