When the Cloud Leaves Earth: Protecting Data Center Innovation in Orbit

Over the past year, efforts to put data centers into Earth’s orbit have moved from theoretical experiments to funded engineering programs. For example, Google’s Project Suncatcher aims to link solar-powered satellites carrying its own Tensor Processing Units into an orbital AI cluster, and two prototype spacecrafts built by Planet Labs are targeted for an early-2027 launch.[1] Starcloud has already flown an Nvidia H100 in orbit, trained a small language model there, and even filed with the FCC for approval for an 88,000-satellite constellation for data processing.[2] Elon Musk’s SpaceX has also filed for FCC approval[3], and some commentators suggest that space-based compute could eventually undercut terrestrial data centers on cost. Other entrants like Aetherflux (now Cowboy Space Corporation), Axiom Space, NTT, Ramon.Space, and others are rushing to position themselves in the same emerging (outer) space.[4]

The simplified pitch goes like this: no land acquisition, no homeowner neighbors, no municipal power fights, nearly continuous solar power, and the use of space for heat dissipation. Real technological and regulatory challenges still exist, but the attraction is easy to understand, particularly with increasing struggles and rising opposition to terrestrial data centers. Setting aside the vertical and pre-existing IP in the broader field of space exploration and satellite operation, new technologies are emerging that aim to solve the unique engineering challenges to enable data center operation in space at scale, and past investment in patenting is bearing fruit.

For example, Voyager Technologies announced earlier this year that it received a patent for manufacturing orbital optical communications.[5] CalTech’s broadly titled “Space-Based Data Centers” patent officially issued earlier this month as U.S. Pat. No. 12,679,564 from a provisional application filed in 2023. And giants like Intel no doubt have been active in protecting their innovations—the company just announced a new processor called “Starfire.”[6] Companies that wait for the engineering to fully mature before addressing IP will find previously available whitespace already claimed by the time they arrive. And, as with similar leading edge technologies like quantum computing before it, the firms building durable portfolios now are the ones anchoring claims to specific, tested engineering solutions rather than to the broader concept of “a data center in space.”

Bold Vision, Complex Engineering, and Threading the Patentability Needle

As with much in tech, marketing runs the risk of being well-ahead of the required hardware. Space deployment of data centers, especially in the way we currently think of them, carries many real challenges to address. As one example, radiation tolerance for components in space is a non-trivial constraint. While some companies, including Google, report their components can withstand the radiation levels expected over a five-year low-Earth-orbit mission in testing, large scale commodification must follow and be viable. Thermal management is another known problem to address in space such that data center power density, rack architecture, and radiator geometry in orbit will look nothing like a hyperscale data center on the ground.

On-orbit maintenance remains essentially unsolved, and there is not yet an analog to a technician walking into a data hall to swap a failed component. There are efforts to bridge that gap, with companies like Northrop Grumman leveraging DARPA technology to develop on-orbit servicing robots.[7] Still, terrestrial data centers are built around the assumption of continual hardware servicing and refresh cycles; an orbital cluster has to either engineer around failure (redundancy, graceful degradation, over-provisioning) or accept a shorter useful life than a ground-based facility.

None of these challenges render the concept implausible or mean that patent activity is not ripe for action. Rather, it reinforces that durable patent claims in this technology sphere will be the ones tied to specific problems and ways to solve them. Examples include, but are not limited to, radiation-tolerant chip packaging and shielding, heat management and thermal-interface design, intersatellite networking architectures, and navigation. Robust disclosures, targeted claims, and building a durable and mineable portfolio are all needed in a hot, albeit crowded, field with many technologies in play. Thinking about adjacent white space during the design and disclosure phase will only help future portfolio development.

Government Funding and the Strings Attached

There is not yet a quantum computing-style flagship federal incentive program aimed at orbital data centers. But public funding is present in adjacent and increasingly overlapping forms.

Currently, NASA, the Space Force’s Space Systems Command, and DARPA award funding for related technologies. And just like prior awards, any grants or larger funding pushes like ARPA-E’s Genesis Mission could trigger Bayh-Dole disclosure and election obligations, march-in provisions, or government-purpose rights, depending on the specific funding instrument and technology. Growing companies and counsel alike must keep these considerations in mind. As orbital data center technology matures and gets closer to launch, these grants and federal funding considerations may ramp up as well.

The practical guidance doesn’t change from other deep-tech sectors: identify every government-support statement before a patent application is filed, confirm whether the claimed invention was conceived or first reduced to practice under a funding agreement, and build Bayh-Dole compliance into the docketing and prosecution process from the outset rather than discovering a gap during follow-on financing or acquisition diligence.

The Hybrid Strategy Advantage: Patents Where They Add Value, Trade Secrets Where They Provide an Edge

As with many fields, the right answer is very rarely “patent everything,” although disruptive technologies generally result in a flood of filings. But real technological advances that companies can and then do keep secret can strike an attractive balance. Engineering advances like specific manufacturing tolerances, proprietary thermal-interface or material formulations, navigation and positioning algorithms, and calibration data developed through in-orbit testing are all strong trade secret candidates.[8] That is because they are difficult to reverse-engineer from a finished satellite, and valuable precisely because a competitor cannot observe them by inspecting the product from the ground. Trade secret programs must be intentional, and counsel should consider their structure and implementation.

Patents, on the other hand, provide disclosure and visibility to the public. This can enhance marketing and build a fence around strong enforceable rights. Having or at least considering a multi-pronged strategy can be important to mature operating companies and new entrants alike.

Given how many of these programs are likely to be structured as multi-party collaborations where space meets data center processing, agreements handling trade secrecy (and inventorship and ownership of novel advances) deserve a very close look. Joint development agreements and material transfer agreements should be reviewed in each instance for specific IP provisions and should not be left as an afterthought. Teaming and collaboration agreements should allocate prosecution control, licensing rights, and enforcement authority explicitly at the outset, and negotiations captured along the way in memorandums of understanding. Risks related to lopsided agreements sometimes make themselves known years later, after the parties have time to slow down their rapid development cycles. Taking some time at the outset can make sure each party understands what it is agreeing to, and can avoid difficult situations later in the relationship.

Conclusion

Orbital data centers are likely years off from a mature commercial category. But the accelerating advances and opportunity for current investment dictate that IP decisions made now matter more, not less. Companies that tether a patent strategy to real, tested engineering solutions rather than the abstract concept of compute in orbit, build Bayh-Dole and other regulatory compliance into their process from day one, and make deliberate choices about what belongs in a patent application versus what belongs in a lab notebook marked Trade Secret, will be the ones positioned to capture value once the engineering answers are confirmed through launch.

[1] Juby Babu & Jaspreet Singh, Google in Talks with SpaceX for Suncatcher Orbital Data Center Project, Reuters (May 12, 2026), https://www.reuters.com/science/google-spacex-talks-explore-data-centers-orbit-wsj-reports-2026-05-12/; Joey Klender, Planet and Google Explore Computing in Orbit, Space Explored (Jan. 5, 2026), https://spaceexplored.com/2026/01/05/planet-and-google-explore-computing-in-orbit/.

[2] Blake Crosley, The Orbital Data Center Race: Every Major Player, Timeline, and Economic Reality in 2026, Introl Blog (Feb. 21, 2026), https://introl.com/blog/orbital-data-centers-space-computing-race-2026.

[3] Id.

[4] Data Center Dynamics, supra note 1 (describing Axiom Space, NTT, Ramon.Space, Sophia Space, and Aetherflux/Cowboy Space Corporation as additional entrants pursuing orbital data centers).

[5] Voyager Techs., Voyager Secures Breakthrough Patent for Orbital Optical Communications Manufacturing (Jan. 14, 2026), https://voyagertechnologies.com/press-releases/voyager-secures-breakthrough-patent-for-orbital-optical-communications-manufacturing/.

[6] https://spacenews.com/intel-debuts-new-chip-for-space-computing/.

[7] https://www.northropgrumman.com/what-we-do/space/space-logistics-services/space-industrial-revolution.

[8] See prior discussion of space IP, here: The Space Industry’s Legal Landscape: Developing Robust Intellectual Property Portfolios.