What Policies Do Orbital Data Centers Need Before They Can Scale?

Key Takeaways

• Orbital data centers need evidence of useful performance, safe operations, and paying customers.
• Faster licensing can support experiments without settling whether large networks are economical.
• Environmental benefits depend on manufacturing, launch, replacement, and disposal across a mission.

Orbital Data Centers Enter the Policy Debate

On September 28, 2026, the Information Technology and Innovation Foundation (ITIF) published Policy Considerations for Orbital Data Centers. Written by Ellis Scherer and Elizabeth Sanchez, the paper recommends changes involving satellite licensing and spectrum access. It also addresses cybersecurity, launch infrastructure, and orbital sustainability.

Those recommendations are proposals from a policy research organization. Their publication does not grant operating authority, allocate radio frequencies, or establish a government funding program. That distinction matters because an announcement about policy can otherwise sound like an announcement that a commercial system has cleared its regulatory requirements.

An orbital data center places computing or storage equipment aboard a spacecraft. The phrase covers substantially different designs, from equipment serving a particular space mission to proposed satellite networks intended to handle demanding computing tasks for customers on Earth. Treating these designs as interchangeable makes their commercial prospects harder to assess.

The September paper argues that selected space-based workloads offer a more plausible near-term application than wholesale replacement of terrestrial infrastructure. Its central policy position is that governments should permit useful experimentation without presuming that the technology will become competitive at large scale. This leaves the commercial outcome open rather than treating regulatory reform as evidence of future profitability.

The policy debate has already attracted government scrutiny. In an April 28, 2026 assessment, the U.S. Government Accountability Office described both potential resource savings and unresolved engineering barriers. It distinguished smaller facilities processing information generated in space from larger systems proposed for training artificial intelligence models.

For public decision-making, that distinction suggests a practical starting point: evaluate the service being offered before evaluating the size of the announced network. A satellite that processes observations before transmission presents a different investment case from a constellation expected to compete with established cloud providers. Each requires evidence appropriate to its function.

The distinction between orbital computing business models also changes what counts as progress. A successful onboard calculation can establish that equipment works under particular conditions. It cannot, by itself, establish that customers will purchase enough computing capacity to recover the cost of an entire network.

Governments face a related measurement problem. Approval speed is relatively easy to count, but public value depends on what an approved system actually delivers and what obligations it creates for other space users. A useful policy framework would track both: the time required to obtain a decision and the quality of the operational evidence available when that decision is made.

Processing Information Near Its Point of Collection

The European Space Agency announced on July 15, 2025, that Φsat-2 had entered its science phase. The small satellite had completed commissioning and was delivering data using onboard algorithms that process Earth observation imagery. Its applications include identifying cloud-obscured images and detecting features associated with disasters.

This is a concrete demonstration of computing near the instrument producing the information. It provides a more useful reference for evaluating selected orbital services than a rendering of a future server constellation. The mission demonstrates particular capabilities; it does not establish the economics of a general-purpose cloud business.

The underlying mechanism is straightforward. A sensor collects information, a processor evaluates it, and the spacecraft transmits selected results or compressed data to the ground. Downlink means that space-to-ground transmission. Reducing the volume requiring transmission can make a limited communications connection more productive.

However, filtering introduces choices about what to retain. Information discarded aboard a spacecraft may later prove valuable for a different scientific question or an improved processing method. A customer purchasing processed observations should understand whether the original measurements remain available and how onboard decisions can be audited.

The Φsat-2 experience illustrates this concern. The agency explained that its team expanded the available data products to support application development and training, rather than distributing only final processed results. That decision connects computing architecture to the needs of the people who will use the information.

A proposed orbital computing hub takes a different approach by separating some processing capacity from individual sensing spacecraft. Such an arrangement could let multiple missions share equipment. Its value would depend on reliable connections between those missions and the hub, together with agreements governing access and service continuity.

Centralizing processing also creates a commercial dependency. A spacecraft operator would need to know whether capacity remains available during periods of heavy demand and what happens if the computing provider fails. Those concerns belong in procurement terms and system design before a customer commits to an architecture that cannot easily change after launch.

The service being purchased should determine the performance test. For an Earth observation application, an appropriate test might measure the time between collecting an image and delivering a verified result. For a storage service, the relevant test could involve successful retrieval after a communications interruption. Processor specifications alone cannot answer either question.

Policy can encourage this form of measurement through demonstration contracts. A contract tied to a defined user outcome would reveal more than a grant conditioned only on placing hardware in orbit. Publishing enough information to assess results would also help distinguish a repeatable service from a successful but narrowly configured experiment.

Heat and Radiation Set Physical Limits

Vacuum does not provide a surrounding flow of cold air that carries heat away from a computer. The National Aeronautics and Space Administration’s thermal control guidance explains that spacecraft exchange heat with their external environment through radiation. Internal components can transfer heat through physical connections, but the spacecraft still needs a way to emit the resulting waste heat.

For orbital data centers, this makes cooling equipment part of the computing system’s basic architecture. A processor’s electrical demand and its heat output must be considered together. Adding computing capacity can require changes to the surfaces and equipment responsible for rejecting heat.

That relationship complicates simple claims about plentiful solar energy. More available power is useful only if the equipment can turn it into sustained work without exceeding operating temperatures. A design could have sufficient electrical generation and still face a thermal constraint that prevents its processors from running continuously at their advertised performance.

Performance reporting should reflect this distinction. Short demonstrations can establish that a calculation completes successfully, but buyers also need evidence about sustained operation. Tests should state the environmental conditions and identify any periods when equipment reduced its activity to remain within temperature limits.

Radiation creates a separate set of concerns. NASA’s spacecraft avionics assessment discusses radiation mitigation and reliability in onboard electronics. Depending on the mission and hardware, protective measures can include selecting appropriate components and designing systems to detect or recover from faults.

For a computing customer, recovery behavior can matter as much as the initial failure rate. A system that detects corrupted work and restarts it may preserve accuracy at the cost of additional time and energy. A system that returns an incorrect result without detecting the problem presents a different risk.

These distinctions suggest that qualification should include the output of the service, not just survival of its components. Successful operation requires confidence in the calculation and the procedures used to recover after disruption. The supporting evidence should make clear which failure conditions were tested and which remain assumptions.

Google’s September 24, 2026 Project Suncatcher update describes an early effort to test computing hardware under space conditions. Its discussion includes radiation and cooling concerns. The update supports treating the project as an engineering investigation, rather than proof that a large commercial orbital computing service is already available.

Coverage of accelerated computing in space helps explain why processor developers and spacecraft businesses are pursuing these tests. Policy should leave room for different technical approaches, since prescribing a particular hardware design too early could exclude methods that achieve the same safety or reliability objective more effectively.

Communications Permission Is Part of the Product

A computing service cannot deliver value if customers cannot reliably send it work or retrieve results. For orbital data centers, communications is part of the service architecture, even when promotional material concentrates on processors or solar panels. The commercial product includes a usable connection between the customer and the equipment.

The International Telecommunication Union (ITU) describes satellite regulation and coordination as an international process involving national administrations. Governments submit information about planned networks and coordinate frequency use to address harmful interference. International coordination and domestic authorization serve related but distinct purposes.

This framework means that an orbital computing proposal must identify how its radio links fit into existing arrangements. A description such as “data center” does not itself specify the communications service or the frequencies involved. Regulators need technical information about the proposed transmissions and their relationship to other users.

Predictability could improve without removing those questions. Clear application instructions and early identification of missing information would help developers understand the evidence required. Agencies could also explain which elements of a hybrid computing and communications system fall within their authority, reducing uncertainty before equipment design becomes expensive to change.

Optical communications, which use light rather than conventional radio transmissions for data transfer, offer another design option. Their presence does not settle every network requirement. A system may still depend on other communications channels for spacecraft control or for connections outside the optical network.

An assessment of a proposed service should distinguish its internal connections from its route to the customer. Fast exchanges between nearby satellites do not automatically establish the same performance between orbit and a terrestrial customer’s systems. The entire route needs to support the promised service.

The developing orbital computing supply chain includes businesses with different relationships to communications infrastructure. Some proposals combine computing with connectivity; others rely on partners. That difference affects which company controls service quality and which agreements must remain in force for the computing business to function.

For procurement, this raises an allocation-of-responsibility question. Customers should know whether one provider is accountable for the complete service or whether they must coordinate separate spacecraft and communications contracts. An outage that falls between those contracts can leave a buyer with functioning equipment but no usable service.

International coordination also affects market access. A company’s ability to operate one element of a network does not automatically establish permission for every ground connection it wants to use. Commercial planning should treat the geographic reach of authorized communications as part of the product specification, alongside capacity and price.

Cybersecurity and Jurisdiction Extend Beyond the Satellite

An orbital computer does not leave legal responsibility behind when it leaves Earth. Under the Outer Space Treaty, nongovernmental space activities require authorization and continuing supervision by the appropriate state party. The treaty also provides that the state of registry retains jurisdiction and control over its registered space object.

Those provisions establish an international framework for responsibility. They do not answer every question about a customer’s information or every contract involving a space-based service. A company evaluating an orbital provider still needs to identify the relevant operators and the legal arrangements governing the service.

The distinction between controlling the spacecraft and controlling customer data deserves explicit treatment. A spacecraft operator may need authority to protect vehicle safety without having unrestricted access to stored information. Contracts and technical permissions should reflect that separation rather than assuming that one broad administrative account can safely serve every purpose.

The U.S. National Institute of Standards and Technology published satellite command-and-control guidance in 2022. It applies the Cybersecurity Framework to the ground segment, emphasizing the command and control of satellite platforms and payloads. This is guidance for managing risk, not an automatic certification of any orbital computing business.

The ground segment includes the equipment and systems on Earth that support a mission. An orbital service can depend on these facilities even if its computing hardware operates far above them. Protecting the satellite but neglecting the systems that issue commands would leave part of the operating chain exposed.

A useful security assessment would follow the chain of authority. It would identify who can upload software and who can alter access permissions. It would also examine how the operator would respond if trusted credentials were stolen, since a technically valid command can still be unauthorized in a commercial or organizational sense.

Incident response requires advance preparation. A provider should establish how it can isolate a customer workload without interfering with spacecraft safety. Customers, in turn, need procedures for withdrawing sensitive work or retrieving their information if the provider suspends normal operations.

Data sovereignty, meaning authority over data and its handling, is sometimes presented as a benefit of moving infrastructure into space. Ownership and operational control remain more informative than physical altitude alone. A service’s dependencies can include software suppliers and communications providers located in different jurisdictions.

This creates a practical test for sovereignty claims. The provider should explain who can operate the service independently and which dependencies cannot be substituted. A promise of national control carries less weight if essential operating permissions or software access remain outside that control.

Public buyers could improve consistency by stating security outcomes in their contracts. Requirements for verified updates and recoverable operations would be more useful than a general demand that a system be “secure.” The evidence should demonstrate how the provider meets those outcomes throughout the service’s operating life.

Environmental Claims Need a Complete Accounting

Thales Alenia Space published results from the ASCEND feasibility study on June 27, 2024. The European Commission funded the work through Horizon Europe. The study examined the technical and environmental feasibility of space-based data centers.

Its environmental conclusion came with a substantial condition. Thales Alenia Space reported that a significant reduction in emissions would require development of a launcher with lifecycle emissions one-tenth of those used as the study’s comparison. A conditional study result should not be presented as a measured environmental benefit from an operating facility.

The distinction is important for procurement and public communication. A design can reduce one category of resource use and increase another. Avoiding water consumption at a terrestrial computing location does not establish the overall environmental performance of equipment that must be manufactured and launched.

Lifecycle accounting examines impacts across the system’s existence, including production and retirement. For orbital computing, that accounting should also identify how replacement hardware enters the calculation. A system with a short useful life may require a different launch and manufacturing schedule from one designed to operate longer.

The unit of comparison matters just as much as the boundary of the assessment. Comparing nominal computing capacity can conceal differences in availability or performance. A stronger comparison would measure a defined amount of completed work at a stated level of accuracy and service reliability.

Replacement assumptions deserve particular attention because computing hardware and spacecraft can age on different schedules. A functioning vehicle might carry processors that no longer meet customers’ needs. Conversely, a capable processor cannot generate revenue if another spacecraft subsystem ends the mission.

Orbital safety introduces costs that do not appear in a simple electricity comparison. The Government Accountability Office’s April assessment identifies congestion and collision risks associated with expanding satellite populations. It also notes uncertainty concerning servicing and the implications of more frequent retirement.

Those concerns support requiring a credible retirement plan before deployment. The plan should explain how normal end-of-life operations will proceed and what happens if the vehicle loses some capabilities earlier than expected. Financial arrangements should identify who retains responsibility if the original business no longer operates.

A sustainability claim also needs a date and a baseline. Terrestrial electricity supplies and computing technology can change during the development of a space system. A comparison based on an outdated ground facility could overstate the relative benefit of a future orbital design.

Public authorities could make environmental evidence more useful by asking applicants to disclose the assumptions that most influence the result. Independent assessment could then focus on those assumptions rather than debating a single headline number. The objective would be a comparison that can be revised when measured operating data becomes available.

Commercial Viability Depends on Delivered Service

A processor’s purchase price is only one component of an orbital computing business. Investors must also account for the spacecraft and the means of delivering its service. An assessment of terrestrial and orbital computing costs is most useful when it compares complete systems rather than isolated hardware.

The economic denominator should be useful work delivered to paying customers. Installed capacity can remain idle because demand is low or because another part of the system limits operations. A financial model that assumes continuous paid use should identify the contracts or demand evidence supporting that assumption.

Customer acquisition may be as consequential as engineering performance. A technically successful platform still needs organizations willing to adapt their workflows and accept its service conditions. Buyers may require testing and integration before shifting work that already runs satisfactorily elsewhere.

Space-based customers could have reasons to accept a different price structure from terrestrial cloud users. Processing that avoids an otherwise expensive transmission may offer value even if the processor itself costs more to operate. The relevant comparison is the cost and performance of the complete mission workflow.

This does not eliminate competition. An onboard processor may compete with a shared orbital service, and either may compete with improvements in downlink capacity. The customer’s preferred approach can change as these alternatives improve.

Insurance and contractual responsibility also affect delivered cost. A provider should explain what happens to customer commitments after a launch failure or an early equipment loss. Compensation terms and backup arrangements influence the value of a service promise even when they do not change the spacecraft’s technical capabilities.

Hardware renewal creates another financing question. Revenue from an operating system may need to support its replacement before that system reaches physical failure. A business plan that treats replacement as an optional future expansion risks confusing current cash generation with sustainable profitability.

Government procurement can help establish demand, but the contract’s design matters. Purchasing a defined service would test whether the provider can meet a customer need. Funding equipment development tests a different proposition and should be described accordingly.

Competitive procurement could leave room for terrestrial or onboard alternatives when the required outcome permits them. If an agency needs computation in orbit, it should explain the operational reason. If the need can be met from Earth, restricting competition to an orbital solution could obscure the cost of that choice.

Public investment decisions should also separate shared infrastructure from company-specific support. Launch facilities may serve multiple users, but that does not make every proposed expansion economical. Funding decisions need evidence of demand and a clear account of who bears the risk if anticipated activity does not materialize.

The Next Decisions Should Produce Comparable Evidence

The most useful next milestone would be a demonstration that connects technical performance to a customer outcome. A public announcement that equipment reached orbit establishes deployment. A demonstration that repeatedly meets a defined service requirement provides stronger evidence about whether a business can develop around it.

For computing trials, reporting should explain the workload and the conditions under which it ran. The information should distinguish sustained operation from short tests. It should also identify the effort required to recover from interruptions rather than reporting only successful runs.

Communications tests need similarly clear boundaries. Results should identify whether measurements concern a connection between spacecraft or a complete path to a customer. Without that distinction, a high internal transfer rate can be mistaken for the performance available to the user purchasing the service.

Licensing decisions offer another set of milestones. An application, a request for additional information, and an authorization represent different stages. Coverage of orbital data centers should identify which stage has actually occurred and whether an authorization contains conditions limiting deployment or operations.

Environmental evidence should move from design assumptions toward measured inputs as projects mature. Manufacturing data and actual hardware lifetimes would permit more informative comparisons. A developer should be able to revise an early claim if operating experience changes the assumptions that supported it.

Customers will also need evidence of continuity. A trial that depends on a small engineering team manually resolving every problem can still be valuable research. It does not establish that the same system can provide routine service at the staffing level assumed in its financial model.

For governments, staged commitments could match public exposure to demonstrated progress. A limited experiment could establish technical evidence before a larger purchasing decision. Each stage should have a defined purpose so that continued funding reflects new information rather than the cost already incurred.

Global participation makes consistent terminology useful. “Demonstrated,” “authorized,” and “commercially available” should describe different achievements. Using them precisely would make announcements from different jurisdictions easier to assess without requiring identical regulatory systems.

There is also a role for independent testing. Customers and public authorities would benefit from performance assessments that state limitations as clearly as achievements. Access to enough methodological detail could allow another qualified evaluator to understand what a result proves.

The September policy discussion is most productive when it leads to these more specific decisions. Permission to experiment can expand the available evidence, but the evidence must still determine which applications justify further deployment. A failed commercial hypothesis can be an informative result if the experiment was bounded and its findings are usable.

Summary

Orbital data centers could create a new dependency between digital services and spacecraft operations. That possibility gives customers a reason to consider exit arrangements before committing to a provider. Data retrieval and workload portability would matter even if the service performs well during its initial contract.

Public policy can support experimentation by making responsibilities and decision procedures understandable. Commercial scale should follow evidence that a defined service works, attracts demand, and meets its obligations over time.

The unresolved question is broader than whether computers can operate in space. It concerns whether organizations can buy dependable computing services under terms that remain workable when equipment fails, ownership changes, or the original business plan proves wrong.

Appendix: Useful Books Available on Amazon

• Spacecraft Systems Engineering
• Designing Data-Intensive Applications
• Handbook of Space Law
• The Space Economy
• The Space Business

Appendix: Top Questions Answered in This Article

What are orbital data centers?

Orbital data centers place computing or storage equipment aboard spacecraft. They can serve individual missions or form part of a network providing shared services. The term covers different architectures, so a proposal for a large computing constellation should be assessed separately from a small onboard processing demonstration.

Did the September 28, 2026 report establish new regulations?

No. The Information Technology and Innovation Foundation publication presents policy recommendations concerning orbital computing. Publication of the paper does not authorize a spacecraft or create a funding program. Any government action responding to its recommendations would require its own decision-making process and documentation.

What computing applications have already been demonstrated in orbit?

The European Space Agency’s Φsat-2 mission demonstrates onboard processing of Earth observation imagery. In July 2025, the agency announced completion of commissioning and delivery of science data. This supports the practicality of selected onboard applications, but it does not establish the economics of a general-purpose orbital cloud.

Why is cooling difficult in space?

Vacuum lacks the surrounding air that can carry heat away through convection. A spacecraft must transfer internal heat to surfaces that emit thermal radiation. Providing adequate heat rejection can affect the design and operating limits of an orbital computing system, even when sufficient electrical power is available.

Does solar power make orbital computing inexpensive?

Solar energy does not remove the cost of collecting power or supporting computing equipment. A provider must also pay for the spacecraft and its deployment. The economic comparison depends on useful work delivered over the operating life, including interruptions and replacement requirements.

Why do orbital data centers need spectrum coordination?

Systems using radio communications must account for other users of the same frequencies. International coordination involves national administrations and procedures intended to address harmful interference. An orbital computing label does not replace the technical information and authorizations needed for the communications links supporting the service.

Are orbital computers outside national jurisdiction?

No. The Outer Space Treaty requires authorization and continuing supervision of nongovernmental space activities by the appropriate state party. It also provides for jurisdiction and control by the state of registry over a registered space object. Other questions concerning customer information depend on the relevant legal and contractual arrangements.

Are orbital data centers automatically better for the environment?

No. Environmental performance depends on the full lifecycle and the terrestrial system used for comparison. A credible assessment must consider manufacturing and deployment, together with replacement and retirement. Potential savings in land or water use do not independently establish a lower overall environmental impact.

How should governments evaluate public funding requests?

Governments can distinguish technology development from purchases of an operating service. Each funding decision should specify the evidence it is intended to produce and the obligations attached to support. Staged commitments can help relate additional spending to demonstrated results instead of relying entirely on projected demand.

What evidence would show that orbital computing is commercially viable?

Useful evidence would combine sustained technical performance with paying demand and credible operating costs. Customers would also need dependable communications and clear remedies for service failure. A successful launch or calculation is meaningful progress, but neither alone demonstrates a business capable of financing continued operations and replacement.

Appendix: Glossary of Key Terms

Orbital Data Center

Computing or storage equipment operating aboard a spacecraft, either for a particular mission or as part of a shared service. The phrase can describe very different system sizes. It does not, by itself, establish commercial availability or performance comparable to terrestrial facilities.

Downlink

A communications connection that transfers information from a spacecraft to a receiving facility on Earth. Its available capacity can influence how much information a mission can deliver. Processing data aboard a spacecraft may reduce the amount that needs to pass through that connection.

Thermal Radiation

The emission of energy as electromagnetic radiation because an object has a temperature. Spacecraft use radiating surfaces to reject waste heat to their surroundings. This process differs from cooling through air moving across equipment and must be included in the spacecraft’s design.

Ground Segment

The facilities and systems on Earth that support a space mission. These can provide spacecraft control or handle information transferred between orbit and users. Their security and availability affect the service even when the main computing equipment operates aboard a satellite.

Data Sovereignty

Authority over data and the conditions governing its storage or use. For an orbital service, assessment of sovereignty involves ownership and operational control, together with relevant legal obligations. Moving equipment into space does not independently establish control over every dependency supporting the service.

Lifecycle Accounting

An assessment that considers a system across its existence rather than examining only its operating phase. For orbital computing, this includes equipment production and deployment, followed by replacement and retirement. The selected boundaries and comparison baseline can materially affect the resulting environmental assessment.

Workload Portability

The ability to move a computing task and its associated information between service environments. Portability can help a customer change providers or respond to an interruption. It depends on compatible arrangements and usable data formats, rather than physical access to the original computing hardware.