Blue Origin Project Sunrise: The Race to Build Data Centers in Orbit

Key Takeaways

• Blue Origin filed an FCC application on March 19, 2026 for up to 51,600 orbital data center satellites
• Project Sunrise relies on optical inter-satellite links through the TeraWave network for primary data transfer
• The constellation targets sun-synchronous orbits at 500-1,800 km to harness near-constant solar power

The Ambition Behind Project Sunrise

On March 19, 2026, Blue Origin filed an application with the Federal Communications Commission that, in its sheer scale, signals an entirely new chapter for the company best known for building rockets. The filing, submitted by regulatory counsel Kaitlyn Mahoney and senior regulatory engineer Ryan Henry from Blue Origin’s Kent, Washington headquarters, seeks authority to launch and operate what the company calls the Blue Origin Orbital Data Center System – known internally as Project Sunrise.

The numbers alone set it apart from anything the company has previously attempted. Project Sunrise calls for a non-geostationary satellite orbit, or NGSO, constellation of up to 51,600 satellites operating in sun-synchronous orbits ranging from 500 to 1,800 kilometers in altitude, with orbital inclinations between 97 and 104 degrees. Each orbital plane in the constellation would contain approximately 300 to 1,000 satellites. It’s a plan that dwarfs most comparable proposals, arriving at a moment when the space industry is openly racing to colonize low Earth orbit with computing infrastructure.

That race has a specific driver: the explosion in demand for artificial intelligence computing. Data centers on the ground are struggling to keep pace. Land is scarce in key markets, electrical grids are strained, water supplies used for cooling are under pressure in drought-prone regions, and permitting timelines are long. The International Energy Agency reported that data center and AI-related electricity consumption, which stood at roughly 460 terawatt-hours in 2022, was expected to more than double to over 1,000 terawatt-hours by 2026 – an amount roughly equivalent to Japan’s entire national energy consumption. Against that backdrop, the logic behind putting data centers in orbit is no longer purely speculative.

What Project Sunrise Actually Is

It’s worth being precise about what Blue Origin is proposing, because Project Sunrise is distinct from another constellation the company announced just two months earlier. In January 2026, Blue Origin unveiled TeraWave, a 5,408-satellite network designed to deliver enterprise-grade connectivity at speeds up to six terabits per second. TeraWave is a communications network. Project Sunrise is something else: an actual computing infrastructure deployed in orbit, using satellites as data centers rather than as conduits for transmitting data between Earth-based facilities.

The FCC filing describes Project Sunrise satellites as operating to support “data centers in space.” Their primary communication method won’t be radio frequency transmissions at all. Instead, the satellites will rely on optical inter-satellite links, routing traffic through TeraWave’s mesh backhaul network and other systems to reach the ground. This design choice is significant. Optical links can carry enormous volumes of data without consuming radio spectrum that other operators need, and they’re far harder to intercept than RF signals – a consideration that will matter to government clients. The satellites will carry onboard communications systems capable of telemetry, tracking, and command functions in the Ka-band, specifically 18.8 to 19.3 GHz for space-to-Earth transmissions and 28.6 to 29.1 GHz for Earth-to-space, but only on a non-interference, non-protected basis.

Blue Origin isn’t asking for priority use of that spectrum. The company is explicitly accepting that any other licensed operator takes precedence, and that Project Sunrise satellites will need to avoid causing harmful interference. That positioning is part of a broader regulatory strategy designed to reduce friction with the FCC while the agency works through a pipeline of similar applications from other companies.

Because the constellation is still being designed, Blue Origin has disclosed its plan certifications and design intentions rather than a fully matured debris mitigation plan. The company has requested a waiver of the FCC’s completeness requirement for exactly this reason, noting it will revise the orbital debris mitigation plan as satellite design matures. That’s an honest acknowledgment of where the program actually stands – early stage, but serious enough to require regulatory positioning now.

The Sun-Synchronous Orbit Strategy

The choice of sun-synchronous orbit for Project Sunrise isn’t incidental. At inclinations between 97 and 104 degrees, satellites in these orbits maintain a nearly fixed relationship with the sun, passing over each point on Earth at roughly the same local solar time with each orbit. Certain dawn-dusk sun-synchronous planes receive sunlight almost continuously – a property that transforms the orbital environment from a technical challenge into an asset for power-hungry compute workloads.

A terrestrial data center running AI training clusters draws power continuously from an electrical grid. If that grid uses fossil fuels, the compute has a carbon footprint. If it uses renewables, there’s still the question of whether enough capacity exists when it’s needed. A satellite in a dawn-dusk sun-synchronous orbit doesn’t have these constraints. It faces the sun almost without interruption, converting solar energy to electricity directly. There’s no land acquisition cost, no permitting delay, no grid connection fee, no cooling water requirement. Blue Origin’s application argues that this combination of factors fundamentally changes the economics of compute capacity in a way that benefits not just the company but the broader industry and the communities that have been dealing with the side effects of massive terrestrial data center buildouts.

Loudoun County in Virginia, often called Data Center Alley, hosts more than 250 operational data centers – the largest hub of its kind in the country. The electrical and water demands on that region are substantial. Shifting energy-intensive compute loads to orbit would, in principle, reduce that kind of pressure on specific communities and their infrastructure.

That said, there’s a real question about whether the sun-synchronous orbital bands can absorb the volume of satellites that companies are collectively proposing. Sun-synchronous orbits represent a finite resource. The altitude bands most useful for continuous solar power are relatively narrow, and they’re already the target of multiple competing proposals, including from SpaceX and Starcloud. Managing sharing arrangements among dozens of constellations operating in overlapping orbital shells will be one of the more contentious regulatory and operational challenges facing the FCC and the industry in the years ahead.

How the Constellation Is Designed to Work

Project Sunrise satellites won’t be identical. Blue Origin’s application specifies that the company will develop and deploy multiple hardware versions to address the coverage and operational requirements of different orbital planes. At least three antenna variations are planned, tailored to different parts of the constellation’s coverage geometry. The radio frequency requirements will remain uniform across the entire system – a choice that simplifies ground operations and spectrum coordination.

The constellation’s operational concept relies on optical links as the primary data pathway. Satellites communicate with each other and with Blue Origin’s TeraWave network via laser crosslinks, routing compute traffic through a mesh that eventually reaches terrestrial endpoints via optical ground terminals. Radio frequency systems serve a backup and operational safety role: telemetry, tracking, and command during launch and early orbit, supplemental support during normal operations, and critical communications during end-of-mission deorbit procedures.

This hybrid approach – optical as the workhorse, RF as the safety net – is becoming a pattern in next-generation constellation design. It allows a system to maximize data throughput without making an FCC case for large-scale spectrum allocation, and it dramatically reduces the interference footprint that the constellation presents to other operators. For a company seeking regulatory approval quickly and without triggering a processing round, it’s a smart design choice.

The satellites will be equipped with frequency-agile transmitters capable of dynamically selecting low-bandwidth operating frequencies as needed during TT&C operations. That frequency agility gives the system an additional layer of interference avoidance – rather than parking on a fixed channel, satellites can shift their TT&C transmissions to wherever spectrum is momentarily clear, reducing the chance that any interaction with another operator becomes a persistent problem.

Blue Origin’s Regulatory Strategy

The FCC application for Project Sunrise includes multiple waiver requests, and understanding them reveals as much about the company’s strategy as the technical specifications do.

The first and most structurally important is a waiver of the processing round requirements under 47 C.F.R. §§ 25.155(b) and 25.157(c). Normally, when a company files for NGSO spectrum rights, the FCC initiates a processing round – a defined window during which other applicants can file competing applications and establish their own rights in the same frequency band. This process is designed to prevent spectrum warehousing by ensuring that early filers can’t simply lock out competitors. But it also takes time, and Blue Origin is asking the FCC to skip it.

The argument is specific and legally grounded. Blue Origin isn’t seeking priority spectrum rights. It’s operating on a non-interference, non-protected basis in bands already allocated for NGSO fixed-satellite service. There’s nothing to warehouse. Any other operator can use the same spectrum and take precedence over Project Sunrise. The FCC has granted similar waivers in the past – to Iridium

The second waiver request targets the milestone and surety bond requirements under 47 C.F.R. §§ 25.164 and 25.165. Under standard FCC rules, an operator must launch 50 percent of its proposed constellation within six years of receiving authorization, and the remainder within nine years, with financial penalties escalating from one million to five million dollars for failures to meet those timelines. Blue Origin is asking to be exempt from both the milestones and the bond.

The argument here rests on the FCC’s own pending rulemaking. The agency’s Space Modernization for the 21st Century Notice of Proposed Rulemaking, filed in October 2025, questions whether the existing surety bond structure is still necessary to prevent spectrum warehousing, given other fee structures that have since been put in place. Blue Origin agrees with that analysis and asks that it be applied to Project Sunrise. Since the company isn’t seeking priority spectrum rights anyway, the concern that motivates the milestone requirement – that an applicant might sit on spectrum authorization without deploying, effectively blocking others – simply doesn’t apply.

Two additional waiver requests address technical filing completeness. Because the satellite design is still maturing, Blue Origin can’t yet submit a fully detailed orbital debris mitigation plan. And because the Schedule S web form used for technical submissions can’t practically accommodate the full orbital parameter data for 51,600 satellites across multiple configurations, the company is providing a representative sample – covering the minimum, maximum, and select intermediate orbital parameters – rather than an exhaustive dataset. These are pragmatic acknowledgments of where the program is, not attempts to hide anything.

Taken together, the waiver package is consistent with a company that wants to secure regulatory position early, without overstating the program’s maturity. Whether the FCC accepts all of them remains to be seen, but the legal precedents cited in the application are real.

The Terrestrial Data Center Problem That Drives All of This

To understand why Blue Origin is pursuing Project Sunrise at this scale, and why it’s not alone, the starting point is the terrestrial infrastructure gap that the company’s FCC filing describes.

AI workloads are categorically different from the computing that dominated data centers a decade ago. Training a large language model requires sustained, high-throughput computation over days or weeks, drawing enormous power continuously. Inference – running a model to answer user queries – happens at scale across millions of simultaneous requests. Both functions require not just raw compute capacity, but specific types of processors, low-latency interconnects, and thermal management systems capable of removing the heat those processors generate.

The problem isn’t simply that there aren’t enough servers. It’s that building the infrastructure to support those servers on Earth has become genuinely difficult. Power interconnections in the most desirable data center markets – northern Virginia, the Pacific Northwest, parts of Texas – have waiting lists measured in years. Water-stressed regions are pushing back against new data center construction. Local governments are grappling with the tax and infrastructure implications of facilities that employ relatively few people while consuming enormous amounts of electricity.

A Google feasibility study published in November 2025 attempted to model when space-based data centers might become cost-competitive with terrestrial alternatives. The authors concluded that if launch costs to low Earth orbit reached $200 per kilogram, the economics could work – and they estimated that point might arrive around 2035 if SpaceX’s Starshipprogram scales to 180 launches per year. That timeline is speculative, but the fact that a major technology company produced a formal feasibility study at all reflects how seriously the industry is taking the proposition.

Starcloud, a Y Combinator and Nvidia-backed startup based in Redmond, Washington, has already put hardware in orbit. In November 2025, it launched Starcloud-1, a 60-kilogram satellite carrying an Nvidia H100 processor – the same GPU that powers much of today’s AI training infrastructure on Earth. The satellite was used to run a version of Google’s Gemini AI model in orbit, making Starcloud the first company to train a large language model in space. By February 2026, Starcloud had filed its own FCC application for a constellation of up to 88,000 satellites for orbital data centers. In January 2026, SpaceX filed for authorization to deploy up to one million satellites for a similar purpose.

The wave of applications is accelerating. China has announced a 200,000-satellite constellation focusing on state coordination, data sovereignty, and in-orbit processing for time-critical applications. Axiom Space deployed a space-based data center node to the International Space Station in September 2025. These aren’t fringe proposals; they’re filings from some of the most well-resourced space companies on the planet.

Blue Origin’s Competitive Position

Where does Project Sunrise sit within this crowded field? The answer isn’t entirely clear yet, and that uncertainty is worth acknowledging plainly.

Blue Origin’s application was filed on March 19, 2026, roughly two months after the company’s TeraWave announcement and well after SpaceX and Starcloud had already staked their regulatory positions. The constellation size – up to 51,600 satellites – is dramatically smaller than SpaceX’s one-million-satellite proposal and substantially larger than Starcloud’s 88,000-satellite plan. It sits in an interesting middle ground: large enough to be commercially significant, but not so large as to appear wholly unachievable.

The company’s strongest differentiator may be vertical integration. Blue Origin designs and builds its own rockets. New Glenn, the company’s heavy-lift vehicle, achieved orbit on its maiden flight on January 16, 2025, from Cape Canaveral’s Launch Complex 36. The second flight, NG-2, successfully launched NASA’s ESCAPADE Mars probes on November 13, 2025, and became the first New Glenn mission to land its first-stage booster on a drone ship in the Atlantic. A third mission carrying AST SpaceMobile’s Block 2 BlueBird satellite was targeted for late February 2026. The company is also developing the New Glenn 9×4 variant, a super-heavy configuration with nine BE-4 engines on the first stage and a wider fairing, capable of carrying more than 70,000 kilograms to low Earth orbit, compared to 45,000 for the current design.

Having its own launch vehicle means Blue Origin doesn’t face the same bottleneck that is currently plaguing Amazon Leo(formerly Project Kuiper). Amazon has hundreds of flight-qualified satellites waiting for launch as of early 2026, and has requested an FCC deadline extension because of a lack of available launch capacity. Amazon has signed 27 New Glenn missions for Leo deployment, which raises an immediate conflict-of-interest question that the industry has noticed: Blue Origin, which is owned by Jeff Bezos and will be launching its own Project Sunrise constellation, will also be launching satellites for Amazon’s competing LEO network. Some analysts have speculated that spinning Amazon Leo off to Blue Origin and merging it with TeraWave would ultimately make more sense than running the two operations separately. Neither company has confirmed any such plan.

The TeraWave connectivity network, which provides the backhaul infrastructure that Project Sunrise satellites will depend on for their primary data links, gives Blue Origin an integrated stack that competitors will struggle to replicate quickly. SpaceX can tie its orbital data center satellites to Starlink, and Starcloud’s FCC filing specifically names Starlink, Amazon’s network, and TeraWave as the backhaul systems its constellation will rely on. Blue Origin, by contrast, controls both the compute layer (Project Sunrise) and the connectivity layer (TeraWave) within a single corporate structure.

The Technical Challenges Nobody Should Understate

None of the companies currently filing for orbital data center authorizations have demonstrated that they can operate a data center in space at commercial scale. The proposals represent engineering ambitions that are real and logically coherent, but they also involve solving problems that haven’t been solved yet.

Radiation is among the most persistent. Low Earth orbit doesn’t offer the shielding from cosmic rays and energetic particles that Earth’s atmosphere and magnetic field provide at the surface. Electronics exposed to the LEO radiation environment degrade faster than their terrestrial equivalents. The processors currently used in AI training clusters weren’t designed for this environment. Developing radiation-hardened versions with equivalent performance – without drastically increasing cost – is an open engineering problem. Blue Origin’s December 2025 initial concept for orbital data centers highlighted the need for radiation-hardened electronics capable of surviving the space environment, and job postings associated with the TeraWave program indicated development of proprietary radiation-hardened silicon, though specifics remain internal.

Thermal management is equally demanding. A dense cluster of high-performance processors generates heat. On Earth, that heat is removed using air cooling, liquid cooling, or in some cases immersion cooling systems that use specialized fluids. In orbit, there’s no air. Heat can only be removed through radiation – the gradual emission of infrared energy into the cold of space from radiative panels. Designing satellite-scale thermal radiators that can remove enough heat to keep AI processors running at full speed, without the panels becoming so large that they interfere with solar arrays or structural integrity, is genuinely hard. Starcloud’s long-term plans contemplate spacecraft with solar arrays four kilometers on a side – a scale that exists nowhere in current satellite engineering.

Then there’s the debris question. Blue Origin’s FCC application acknowledges that the orbital debris mitigation plan is still being matured along with the satellite design itself. That’s understandable for a program at this stage, but it will become a central issue as the application moves toward authorization. The Kessler syndrome – a self-sustaining cascade of collisions that makes certain orbital bands permanently unusable – is not a theoretical concern. It’s a documented risk that the FCC, NASA, and the international space community take seriously. Deploying 51,600 satellites into altitudes that are already home to multiple large constellations will require careful coordination and robust deorbit planning.

The ITU compliance section of Blue Origin’s application is notably brief: the company states that it hasn’t yet submitted system information to the International Telecommunication Union for publication, and will do so “when appropriate.” ITU coordination is a lengthy process, and beginning it early tends to establish priority rights. The fact that Blue Origin hasn’t started is either a reflection of the program’s early stage or a deliberate regulatory strategy – or both.

The Broader Market That Project Sunrise Is Entering

The use cases Blue Origin describes for Project Sunrise in its FCC application span several industries where AI-driven applications are creating genuine demand for compute that isn’t tied to a specific physical location.

Healthcare is one. AI systems capable of earlier disease detection and more accessible diagnostics require substantial compute for training and, increasingly, for real-time inference. A space-based data center running these workloads wouldn’t require a terrestrial facility at all – the compute would simply exist in orbit, accessible via TeraWave or other backhaul networks to ground stations distributed globally.

Agricultural monitoring is another. Precision agriculture systems that integrate satellite imagery, weather data, and soil sensor readings to optimize irrigation, fertilizer application, and harvest timing are already commercial. As these systems become more sophisticated, the compute requirements grow. Blue Origin’s constellation of satellites in sun-synchronous orbits – the same orbital regime used by Earth observation satellites – would be well-positioned to process this kind of data close to where it’s collected.

Climate science represents a third area. High-resolution climate modeling requires enormous compute capacity. The models are data-intensive, and the satellites that gather the raw observational data are already in orbit. Moving the processing infrastructure closer to the data source, rather than downlinking everything to ground stations and then processing it in terrestrial facilities, could reduce latency and network costs while expanding modeling resolution.

These are plausible applications, not marketing abstractions. Whether the economics work at the prices Blue Origin will need to charge to recover the cost of developing and deploying 51,600 satellites is a separate, harder question. Space infrastructure is expensive. Even with New Glenn and its successor, the 9×4 variant, the per-kilogram cost of getting hardware into orbit and keeping it there will need to fall substantially before space-based compute can compete with ground-based alternatives on pure cost grounds. The solar power advantage is real, but so are the development, manufacturing, and launch costs.

The FCC Processing Round Question and Regulatory Precedent

Blue Origin’s request to skip the processing round is the most consequential regulatory ask in the Project Sunrise application, and it’s worth examining how the FCC is likely to approach it.

The processing round system was designed in an era when orbital broadband was dominated by a small number of applicants seeking large blocks of spectrum for consumer services. The rule assumes that granting spectrum rights to one applicant necessarily precludes others from accessing the same frequencies – hence the need for a structured, public process to establish relative priority.

Project Sunrise doesn’t fit that model. The company isn’t seeking spectrum priority. It’s asking to use a band that’s already allocated for its purpose, on a subordinate basis, for a narrow, safety-critical function. The FCC has granted this kind of waiver before – in the Iridium, Northrop Grumman, Boeing, and SpaceX Gen2 cases – and the agency’s own pending Space Modernization rulemaking is explicitly revisiting whether the processing round rules make sense for systems like this one.

If the FCC grants the waiver, it sets a precedent that could simplify authorization for the wave of similar applications the agency is now receiving. That might be a good thing for innovation, or it might accelerate the orbital crowding problem before the regulatory framework has caught up. The FCC under Chairman Brendan Carr has historically been more cautious with mega-constellation approvals, often granting licenses in tranches rather than all at once, and the agency’s track record on Project Sunrise will be watched closely by every company with a similar application pending.

The TeraWave Connection

Project Sunrise can’t be understood in isolation from TeraWave, and the two-constellation strategy Blue Origin is now pursuing represents a more integrated approach to orbital infrastructure than any other company is currently attempting.

TeraWave, announced January 21, 2026, and filed with the FCC on the same date, is a 5,408-satellite constellation with 5,280 spacecraft in low Earth orbit and 128 in medium Earth orbit. The LEO satellites use Q- and V-band spectrum for radio frequency links delivering up to 144 gigabits per second, while the MEO satellites use optical links to provide point-to-point capacity of up to six terabits per second. First deployment is targeted for late 2027 using New Glenn rockets.

The design targets enterprise customers – data centers, governments, logistics networks – rather than the mass consumer market that Starlink serves. A maximum customer base of approximately 100,000 has been cited, compared to Starlink’s reported nine million active subscribers as of late 2025. Each customer gets dedicated, reserved capacity rather than sharing a pool with millions of others. That’s a fundamentally different service model, designed for use cases where predictable throughput is non-negotiable.

For Project Sunrise, TeraWave serves as the essential backhaul layer. Orbital data center satellites don’t have much value if there’s no efficient way to move compute results to and from ground-based clients. TeraWave provides that pathway. The optical inter-satellite links within the Project Sunrise constellation move data between computing nodes in orbit; TeraWave connects those nodes to the ground at the throughput levels that commercial and government clients would require. The two systems are designed to operate together, and the FCC application for Project Sunrise explicitly references TeraWave as part of its communication architecture.

Some satellite industry analysts have questioned whether the MEO layer of TeraWave – which one anonymous engineer described as functionally “bent pipe,” relaying signals between the LEO shell and the ground rather than establishing true global end-to-end paths in space – adds latency and technical complexity that the architecture doesn’t strictly need. That’s a fair technical concern, and it’s the kind of design detail that will likely become clearer as TeraWave moves from regulatory filing to hardware development.

What Blue Origin Has Already Built

To assess how seriously to take Project Sunrise, it helps to look at what Blue Origin has actually delivered in recent years, not just what it’s proposed.

New Glenn reached orbit on January 16, 2025 – a decade after development began, later than originally planned, but functional. Its second flight in November 2025 demonstrated successful booster recovery. New Shepard, the company’s suborbital rocket, completed its 38th flight on January 22, 2026 (though Blue Origin subsequently paused tourism launches to redirect resources to lunar efforts). The company’s Blue Ring multi-mission orbital vehicle is preparing for its first commercial mission, carrying the Caracal optical payload for the U.S. Space Force. The Blue Moon Mark 1 lunar lander is undergoing vacuum chamber testing at NASA’s Johnson Space Center ahead of its first robotic mission.

This is a portfolio of hardware that is real and, in most cases, either flying or close to it. Blue Origin is no longer a company that talks about things it might do. That said, none of what it has delivered yet involves large-scale satellite constellation operations. Manufacturing and deploying 51,600 satellites – or even a meaningful fraction of that number – requires industrial-scale production capabilities that Blue Origin has not yet demonstrated. It’s a fair point of uncertainty that the company’s FCC application itself doesn’t resolve.

By comparison, SpaceX had manufactured and launched more than 10,000 Starlink satellites by early 2026, with a production rate that has allowed the company to continually refresh the constellation with newer hardware. That depth of constellation manufacturing experience is one of the most significant structural advantages SpaceX holds, and it will take Blue Origin years to close the gap, assuming it remains committed to doing so.

The Broader Geopolitical Dimension

The orbital data center race isn’t purely commercial. China’s 200,000-satellite constellation proposal, announced concurrently with the flurry of Western filings in early 2026, frames in-orbit data processing explicitly in terms of state coordination and data sovereignty. A government that can process sensitive data in its own orbital infrastructure, without routing it through ground-based networks owned or influenced by adversarial nations, has a meaningful intelligence and operational advantage.

The U.S. Space Force is already interested in space-based compute for exactly these reasons. Blue Origin’s FCC application notes that its constellation will expand AI accessibility for U.S. companies, a framing that connects naturally to national competitiveness arguments that carry weight in Washington. Blue Ring’s first mission involves a space domain awareness payload for the Space Force, suggesting that Blue Origin’s relationship with the defense sector is already active. Project Sunrise, if it advances, would likely attract interest from government clients who see orbital compute as a strategic asset rather than simply a commercial service.

The FCC’s Space Modernization rulemaking, initiated in October 2025, is partly motivated by recognition that the United States needs to enable faster deployment of commercial space infrastructure to remain competitive with China’s accelerating constellation programs. Blue Origin’s regulatory requests in the Project Sunrise application explicitly cite that rulemaking as support for its waiver arguments – a connection that frames the filing as aligned with national policy objectives, not just commercial interest.

Kessler Syndrome and the Long View

The Kessler syndrome scenario deserves more than a passing mention. In the early 1990s, NASA researcher Donald Kessler modeled a scenario in which the density of objects in low Earth orbit reaches a threshold where collisions generate debris, and that debris triggers more collisions, in a cascade that renders certain orbital bands unusable on timescales of decades or centuries. The debris doesn’t have to be catastrophic to be dangerous – paint flecks and bolt fragments at orbital velocities carry the kinetic energy of bullets.

Decades later, the concern is more concrete. Starlink has already shut down nearly 500 satellites during the first half of 2025, guiding them into controlled reentry. The operational lifespan of LEO satellites is generally estimated at five to eight years, meaning constellations require continuous replacement. Multiple constellations operating in overlapping altitude bands – Starlink, Amazon Leo, TeraWave, Project Sunrise, Starcloud, and others – will require levels of coordination that don’t yet exist at the institutional level.

Blue Origin’s application acknowledges this implicitly by noting that Project Sunrise satellites will operate in sun-synchronous orbits from 500 to 1,800 kilometers, and that end-of-mission deorbit procedures are part of the TT&C operational scope. But the details of how deorbit will work, at what cadence satellites will be replaced, and how the company will coordinate with other operators in overlapping orbital zones are questions that remain to be answered as the satellite design matures. The FCC will require those answers before granting a full authorization.

Summary

Blue Origin’s Project Sunrise represents the company’s most ambitious proposal to date and a bet that the orbital environment can serve as viable infrastructure for AI-driven compute. The March 19, 2026 FCC application for up to 51,600 sun-synchronous orbit satellites is a regulatory positioning move as much as a technical declaration – Blue Origin is establishing its place in a queue of similar filings from SpaceX, Starcloud, and others, using a waiver strategy that reduces regulatory friction by accepting subordinate spectrum status.

The proposition’s logic is sound: sun-synchronous orbits offer near-continuous solar power, space eliminates the land, water, and grid constraints that are throttling terrestrial data center growth, and the AI compute market is genuinely running out of room to expand on the ground. The integration with TeraWave gives Blue Origin a connectivity layer that competitors will need years to replicate, and New Glenn’s operational status removes one of the program’s most obvious risk factors.

What remains genuinely unclear – and this is not a diplomatic hedge but an acknowledgment of real uncertainty – is whether the economics of building, launching, and replacing 51,600 orbital data center satellites will actually pencil out within a timeframe that matters commercially. The Google feasibility study’s estimate of around 2035 for cost competitiveness was predicated on Starship reaching a scale of production that hasn’t happened yet. Blue Origin’s costs per kilogram to orbit, even with New Glenn improving over time, will need to fall substantially. The program’s satellite design is still maturing, the debris mitigation plan doesn’t yet exist in final form, and ITU coordination hasn’t started. These aren’t disqualifying conditions, but they’re real ones.

What Blue Origin has demonstrated is that it’s no longer content to be primarily a launch provider. Project Sunrise, taken alongside TeraWave, Blue Ring, Blue Moon, and Orbital Reef, describes a company that is building toward vertical control of the full space infrastructure stack – launch, mobility, communication, compute, and habitation. Whether all of those programs converge into something coherent, or whether the portfolio’s breadth becomes its own liability, will define the next decade of Blue Origin’s trajectory.

Appendix: Glossary of Key Terms

Non-Geostationary Satellite Orbit (NGSO)

An NGSO is any satellite orbit that is not fixed at the geostationary altitude of approximately 35,786 kilometers above the equator. Unlike geostationary satellites, which appear stationary from Earth and serve fixed coverage zones, NGSO satellites move continuously relative to any point on the ground, requiring constellations of many spacecraft to maintain continuous service. Both Project Sunrise and TeraWave are NGSO systems.

Sun-Synchronous Orbit

A sun-synchronous orbit is a specific type of near-polar orbit in which a satellite’s orbital plane rotates at the same rate as Earth orbits the sun, keeping the satellite in a consistent orientation relative to sunlight throughout the year. Satellites in dawn-dusk sun-synchronous orbits pass over the terminator line between day and night on Earth, receiving near-continuous illumination. This makes them particularly attractive for power-intensive applications like orbital data centers.

Ka-band

The Ka-band refers to a range of radio frequencies between approximately 26.5 and 40 gigahertz used in satellite communications. It supports higher data throughput than older Ku-band systems but is more susceptible to rain fade and atmospheric attenuation. Project Sunrise uses Ka-band specifically for telemetry, tracking, and command operations on a non-interference basis, not for primary data transfer.

Telemetry, Tracking, and Command (TT&C)

TT&C refers to the set of radio frequency functions a satellite operator uses to monitor a spacecraft’s health and position, and to send operational commands. Telemetry provides data on the satellite’s systems, tracking determines its orbit, and command allows the ground station to adjust its behavior. TT&C operations are the primary reason Project Sunrise needs any radio frequency spectrum at all, since its data traffic travels via optical links.

Processing Round

An FCC processing round is a structured regulatory procedure in which the agency establishes a cutoff date for competing applications seeking spectrum rights in the same frequency band, allowing multiple applicants to be evaluated simultaneously and in relation to each other. The purpose is to prevent any single applicant from acquiring spectrum rights in a way that effectively blocks others from entering the same band. Blue Origin is seeking a waiver of this requirement for Project Sunrise on the grounds that it is not requesting priority spectrum access.

Surety Bond

In the context of FCC satellite licensing, a surety bond is a financial instrument that an operator must obtain as a guarantee that it will meet its constellation deployment milestones. If the operator fails to launch the required percentage of its constellation within the specified timeframe, the bond can be called, with values ranging from one million to five million dollars under current rules. Blue Origin is requesting a waiver because it is not seeking protected spectrum rights, which removes the spectrum warehousing concern the bond is designed to address.

Optical Inter-Satellite Link (OISL)

An OISL is a laser-based communication link between two satellites in orbit. Rather than using radio frequency transmissions, which require spectrum allocation and can interfere with other systems, OISLs use tightly focused beams of light to transfer data directly between spacecraft. They can carry far more data than equivalent RF links and present effectively no interference footprint to other operators, making them the preferred architecture for high-throughput constellation backbones including Project Sunrise and TeraWave.

Low Earth Orbit, Medium Earth Orbit, and Geostationary Orbit

These terms describe the three primary altitude regimes for operational satellites. Low Earth orbit runs from roughly 200 to 2,000 kilometers, offering low latency and high data throughput but requiring large constellations for continuous coverage. Medium Earth orbit extends from approximately 2,000 to 35,786 kilometers, providing broader individual satellite coverage at the cost of higher latency and radiation exposure. Geostationary orbit sits at exactly 35,786 kilometers, where satellites appear stationary from Earth’s surface but introduce approximately 600 milliseconds of round-trip signal delay.

Kessler Syndrome

The Kessler syndrome is a scenario, first modeled by NASA researcher Donald Kessler in 1978, in which the density of objects in low Earth orbit reaches a point where collisions generate debris that triggers further collisions in a self-sustaining cascade. The concern is not hypothetical; debris tracking agencies monitor tens of thousands of objects in low Earth orbit, and the risk grows as constellation sizes increase. It is a central consideration in evaluating the orbital debris implications of mega-constellations like Project Sunrise.

International Telecommunication Union (ITU)

The ITU is the United Nations agency responsible for coordinating the global use of radio spectrum and satellite orbital positions. National regulators like the FCC derive their spectrum allocations from ITU frameworks, and satellite operators must ultimately submit system information to the ITU for international coordination and publication. Establishing an ITU filing early confers priority rights, and Blue Origin has noted in its Project Sunrise application that it has not yet done so for this constellation.

Fixed-Satellite Service (FSS)

Fixed-satellite service is an ITU radio service category covering satellite communications between fixed ground stations. The Ka-band frequencies that Project Sunrise will use for TT&C are allocated to NGSO FSS operations on a primary basis, meaning operators with FSS licenses in those bands hold priority rights. Project Sunrise is not requesting FSS status; it will operate within those bands on a subordinate, non-interference basis.

Dawn-Dusk Orbit

A dawn-dusk orbit is a specific sun-synchronous orbit configuration in which a satellite’s ground track follows the terminator line between Earth’s sunlit and dark hemispheres. Satellites in this plane receive sunlight almost continuously rather than experiencing the alternating light and shadow that most orbits involve. For power-intensive applications like orbital data centers, a dawn-dusk sun-synchronous orbit can effectively deliver baseload solar power without the large battery storage systems that would otherwise be needed to bridge eclipse periods.

Appendix: Key Players in the Orbital Data Center Market

The proposals currently before the FCC and being actively developed across the space industry represent organizations at very different stages of maturity, with meaningfully different approaches to the orbital compute problem. The table below summarizes key parameters of each effort as of March 20, 2026.

Blue Origin

Blue Origin enters the orbital data center market with two related but distinct programs running concurrently. TeraWave provides the connectivity backbone for enterprise and government clients, while Project Sunrise places the computing infrastructure itself in orbit, relying on TeraWave for ground backhaul. The company’s ownership of New Glenn gives it a launch independence that Amazon Leo, Starcloud, and most other constellation programs currently lack, and its developing New Glenn 9×4 super-heavy variant could further reduce its per-kilogram-to-orbit costs. The integration between the two programs represents a more complete orbital infrastructure stack than any other single commercial operator has proposed.

SpaceX

SpaceX filed with the FCC in late January 2026 for a constellation of up to one million satellites for orbital AI compute, connected via optical inter-satellite links to the existing Starlink broadband network. The proposal would operate in sun-synchronous orbits between 500 and 2,000 kilometers altitude, using the same Ka-band on a non-interference basis that Project Sunrise has requested. SpaceX’s manufacturing depth, with more than 10,000 Starlink satellites already in orbit and a production line capable of very high launch cadences, gives it a structural advantage no other player can currently match. The potential integration with Elon Musk’s xAI company adds a vertically integrated customer-demand angle that has no parallel among competitors.

Starcloud

Starcloud, formerly known as Lumen Orbit and backed by Y Combinator and Nvidia, is the only company to have already demonstrated working AI compute hardware in orbit. Its first satellite, Starcloud-1, launched November 2025 on a SpaceX rideshare mission, carried an Nvidia H100 processor, and successfully ran a version of Google’s Gemini AI model in space. The February 2026 FCC filing for up to 88,000 satellites targets sun-synchronous dusk-dawn orbits between 600 and 850 kilometers, and long-term Starcloud-4 plans envision spacecraft with solar arrays four kilometers across supporting five-gigawatt data centers. Starcloud is the smallest and most commercially nascent of the major players, but it is the only one that currently has hardware in orbit performing actual compute work.

Google

Google has not filed an FCC constellation application for orbital data centers but has been active in feasibility research and industry partnerships. In November 2025, Google published a study arguing that space-based data center economics could become competitive with terrestrial alternatives when launch costs reach $200 per kilogram, projecting that threshold around 2035. Its Project Suncatcher announced plans in November 2025 to launch two test satellites equipped with AI processing chips in 2027, in partnership with Planet Labs. Google’s participation in the orbital compute space is currently that of a technology developer and feasibility researcher rather than a constellation operator.

China

China’s orbital compute ambitions are organized through state programs rather than commercial FCC filings. A 200,000-satellite constellation has been announced, focused on state coordination, data sovereignty, and in-orbit processing for time-critical applications. Programs including Guowang and the commercially framed Qianfan (Thousand Sails) are designed to establish independent Chinese LEO infrastructure over the coming decade. These systems are unlikely to compete directly for Western commercial clients in the near term, but their existence shapes the geopolitical framing of every orbital data center proposal being made by U.S. companies, and they inform the urgency with which the FCC is considering how to enable faster domestic deployment.

Axiom Space

Axiom Space deployed a space-based data center node to the International Space Station in September 2025, making it the first company to operate commercial computing infrastructure aboard an active crewed space facility. The node serves as a demonstration and development platform rather than a commercial-scale data center, but it provides real operational experience with the thermal, radiation, and logistics challenges involved. Axiom’s planned commercial space station, intended to incorporate and succeed ISS modules in the late 2020s, could provide an iterative platform for developing orbital compute hardware in a crewed environment that purely robotic satellite constellations cannot easily replicate.

Appendix: Blue Origin Program Timeline

The following timeline places Project Sunrise within the broader history of Blue Origin’s program development. It covers major milestones from the company’s founding through the Project Sunrise FCC filing on March 19, 2026.

Appendix: Frequently Asked Questions About Space-Based Solar Power and Computing

How much solar power can a satellite generate in low Earth orbit?

Solar irradiance in low Earth orbit averages approximately 1,361 watts per square meter, compared to roughly 1,000 watts per square meter at Earth’s surface under ideal conditions. Satellite solar panels typically convert 20 to 30 percent of that energy to electricity, meaning a one-square-meter panel can generate 270 to 400 watts in direct sunlight. Scaling that to one gigawatt of power output requires a solar array of approximately one square mile at 30 percent efficiency, which is why proposals for large-scale orbital data centers envision either massive individual spacecraft or very large constellations of smaller ones aggregated into meaningful collective capacity.

Why is sun-synchronous orbit particularly effective for continuous power generation?

Most satellites in low Earth orbit experience regular eclipse periods as they pass through Earth’s shadow, typically losing sunlight for 30 to 40 minutes out of each 90-minute orbit. Satellites in dawn-dusk sun-synchronous orbits follow the boundary between Earth’s sunlit and dark hemispheres, remaining in sunlight for a far larger fraction of each orbit and, in optimal configurations, nearly continuously. This near-constant solar illumination allows orbital data center satellites to generate power without large battery systems that would otherwise need to bridge eclipse periods, significantly reducing spacecraft mass and cost.

How is waste heat removed from computers operating in space?

Terrestrial data centers remove processor heat using air cooling, liquid cooling loops, and in some facilities immersion cooling in specialized fluids. In orbit there is no atmosphere, and convective cooling cannot occur, so heat can only be removed through thermal radiation: the emission of infrared energy from specially designed radiator panels. These panels must be large enough to dissipate the heat generated by dense arrays of high-performance processors, and they compete with solar arrays for spacecraft surface area, creating a design constraint that becomes increasingly difficult to manage as compute density increases.

What is radiation hardening and why does it matter for orbital computing?

Radiation hardening refers to the design and manufacturing techniques used to make electronics resistant to damage from high-energy particles in the space environment, including cosmic rays, energetic protons from solar flares, and electrons trapped in Earth’s radiation belts. Standard commercial processors, including the Nvidia H100 GPUs widely used in terrestrial AI training clusters, were not designed for this environment and can experience data corruption or permanent failure from radiation exposure in low Earth orbit. Developing radiation-hardened versions with performance comparable to commercial chips is a significant and still largely unsolved engineering challenge that most orbital data center companies are only beginning to address at scale.

How do optical inter-satellite links transmit data in space?

Optical inter-satellite links use precisely pointed laser beams to transmit data between satellites at the speed of light, covering distances of hundreds or even thousands of kilometers between spacecraft. The transmitting satellite focuses a laser onto the receiving satellite’s aperture, requiring extremely precise pointing and active tracking systems that account for both satellites’ continuous motion through orbit. Optical links can carry far more data per second than radio frequency links of similar power and require no spectrum allocation, but they are disrupted when clouds or atmospheric conditions block the path between a satellite and a ground terminal, which is why systems like Project Sunrise reserve RF links for ground-segment communications and use optical links only between spacecraft.

How does the latency of orbital compute compare to terrestrial data center latency?

A signal traveling from Earth to a satellite in low Earth orbit at 550 kilometers altitude and back incurs a round-trip propagation delay of approximately 7 to 8 milliseconds, before any processing time is added. Terrestrial data center latency over fiber networks can be 1 to 10 milliseconds for short distances but rises substantially over intercontinental routes with multiple routing hops. For applications where compute happens in orbit and results are transmitted down to a user, the relevant propagation component is one-way, roughly 3 to 4 milliseconds for LEO, which is competitive with intercontinental terrestrial fiber but not with local or regional ground-based infrastructure.

When might space-based data centers become cost-competitive with terrestrial alternatives?

Google’s feasibility study published in November 2025 estimated that space-based data centers could become cost-competitive with terrestrial alternatives when launch costs to low Earth orbit reach approximately $200 per kilogram. The study projected that threshold could arrive around 2035 if SpaceX’s Starship scales to approximately 180 launches per year by then. Starlink currently costs Blue Origin approximately $3 million per satellite to put into orbit, and manufacturing costs have fallen from around $1 million per satellite to roughly $500,000 as Starlink production has scaled, illustrating the cost trajectory that orbital compute economics will need to follow.

What happens to orbital data center satellites at the end of their operational lives?

Current FCC and international guidelines require that LEO satellites deorbit within five years of end of mission to limit debris accumulation in populated orbital bands. Satellites in orbits below approximately 600 kilometers will naturally reenter Earth’s atmosphere and burn up within a few years due to atmospheric drag, while those at higher altitudes require active propulsion to lower their orbits for controlled reentry. Designing propulsion systems into data center satellites adds mass and cost, but it is a regulatory requirement, and the question of how quickly obsolete AI hardware in orbit can be replaced at scale presents a commercial sustainability challenge that terrestrial data centers, where equipment can be upgraded in place, do not face.

Why do orbital data centers not simply replace terrestrial data centers?

Latency, bandwidth constraints between Earth and orbit, and the practical limits of moving large volumes of data through atmospheric and ground segment links prevent orbital compute from fully displacing ground-based infrastructure in the near term. Workloads requiring very low latency to end users, including real-time financial transactions, interactive applications, and localized content delivery, are better served by terrestrial or edge computing facilities near the user. Orbital data centers are most competitive for large-scale batch processing of remote-sensing data, extended AI training workloads where some latency tolerance exists, and applications where avoiding terrestrial energy, land, and water constraints justifies the added cost and complexity of space-based infrastructure.

How large would a solar array need to be to power a commercially significant orbital data center?

Generating one gigawatt of power, roughly the output of a large nuclear power station and a meaningful threshold for commercial data center capacity, would require a solar array of approximately one square mile in low Earth orbit at 30 percent panel efficiency. A typical hyperscale terrestrial data center campus consumes between 100 and 500 megawatts, meaning even a single large facility equivalent would require enormous orbital hardware. Starcloud’s long-term Starcloud-4 concept envisions spacecraft with solar arrays four kilometers on a side to support a five-gigawatt data center, a scale of orbital construction that has no current precedent and would require on-orbit assembly capabilities the industry is only beginning to develop.

Appendix: FCC Regulatory Process for NGSO Constellations

What is the FCC’s role in authorizing satellite constellations?

The Federal Communications Commission has jurisdiction over the use of radio frequency spectrum in the United States and the authorization of satellites operated by U.S. entities. Before any company can deploy a satellite constellation that uses radio frequencies, including for telemetry, tracking, and command, it must receive an FCC license. This applies even to systems like Project Sunrise that rely primarily on optical communications, because any use of radio frequencies in U.S.-regulated spectrum requires formal authorization regardless of how limited or subordinate that use may be.

What does an FCC satellite application include?

A complete NGSO satellite application includes a narrative describing the system’s public interest benefits, a technical annex with detailed orbital and frequency parameters, a Schedule S form containing standardized technical data, and an ownership disclosure exhibit. The application must demonstrate that the constellation will operate within its allocated spectrum, won’t cause harmful interference to other authorized users, and meets the Commission’s rules for orbital debris mitigation. Applications for novel or large-scale systems frequently include waiver requests when standard rules don’t cleanly accommodate the proposed system’s design.

What is a processing round and why does it exist?

A processing round is a structured FCC review process in which one application triggers a public notice and a defined filing window, during which competing applicants can submit their own applications for simultaneous consideration. The process is designed to prevent spectrum warehousing, where a company files for spectrum rights without intending to use them promptly, effectively blocking other operators from accessing the same frequencies. All applications received within the window are evaluated relative to each other, establishing a sharing framework before any single operator receives exclusive priority rights.

When can the FCC waive the processing round requirement?

The FCC has granted waivers of processing round requirements in cases where the applicant is not seeking spectrum priority and where granting the authorization would not preclude other operators from accessing the same frequency band. Precedents include authorizations granted to Iridium in 2016, Northrop Grumman Space and Mission Systems in 2009, and Space Imaging in 2005. In each case, the operator accepted non-interference, non-protected status in its allocated band, removing the spectrum warehousing concern the processing round rule is designed to prevent.

What are milestone requirements and what purpose do they serve?

Under FCC rules at 47 C.F.R. § 25.164, an authorized NGSO operator must launch at least 50 percent of its approved constellation within six years of receiving its license, and the remainder within nine years. These milestones exist to ensure that spectrum rights don’t sit idle for extended periods while the licensee plans or funds its deployment, which would effectively block other operators from using the same spectrum. Failure to meet the milestones triggers surety bond penalties, escalating from one million dollars at six years to five million at nine years, under the current rules that the FCC’s Space Modernization rulemaking is now reconsidering.

What is ITU coordination and why does it matter for international operations?

The International Telecommunication Union maintains the international framework for radio spectrum use and satellite orbital slot assignments. Operators must submit satellite system information to the ITU through their national administration, which in the United States is the FCC, for coordination with other countries’ systems and publication in international databases. Establishing an ITU filing early gives an operator international priority rights against later filers in the same frequency bands and orbital positions, which is why most commercially serious constellation programs initiate ITU coordination as early as possible. Blue Origin has stated in its Project Sunrise application that it has not yet submitted ITU information for this constellation.

How long does FCC satellite authorization typically take?

Authorization timelines vary significantly depending on the complexity of the application, whether waiver requests are included, whether competing applications exist, and the FCC’s current processing bandwidth for satellite filings. Straightforward applications with no contested elements can be processed in six to twelve months. Applications with waiver requests, competing filers, or unresolved public interest questions can take two years or longer. Blue Origin’s multiple waiver requests in the Project Sunrise application introduce uncertainty into the timeline, though the FCC’s stated intent to modernize its space licensing procedures may work in the company’s favor.

What happens after the FCC grants an authorization?

An authorization grants the operator the legal right to launch and operate its constellation using the specified spectrum, subject to conditions the FCC attaches to the grant. Those conditions typically include compliance with the orbital debris mitigation plan submitted with the application, adherence to any deployment milestone schedules, and ongoing coordination with other operators to avoid harmful interference. If the operator wishes to change its constellation’s design, orbital parameters, or spectrum use in material ways after receiving authorization, it must file for a license modification and may need to go through a new review process.

What is the FCC’s current approach to overseeing mega-constellations?

The FCC under Chairman Brendan Carr has generally supported commercial space development while maintaining caution about orbital congestion and debris risks. The agency has historically granted authorizations in tranches rather than for a full proposed constellation in a single grant, allowing it to monitor collision avoidance practices and debris mitigation compliance before authorizing further deployment phases. The Commission’s Space Modernization for the 21st Century Notice of Proposed Rulemaking, released October 29, 2025, proposes updates to processing round rules, milestone requirements, and other procedures to better accommodate the pace and scale of modern constellation programs.

How does the FCC coordinate with other agencies on satellite authorization decisions?

The FCC coordinates with NASA on orbital debris and spectrum compatibility considerations, with the National Telecommunications and Information Administration on spectrum policy matters affecting federal users, and through the State Department with international partners on ITU coordination. For constellations with national security relevance, the FCC also coordinates with the Department of Defense and the U.S. Space Force. Proposed operations that could affect spectrum bands shared with government systems require interagency review, which can extend the timeline for applications touching those bands.

Appendix: Top 10 Questions Answered in This Article

What is Blue Origin’s Project Sunrise?

Project Sunrise is Blue Origin’s FCC-filed proposal, submitted March 19, 2026, to deploy up to 51,600 satellites in sun-synchronous low Earth orbit to function as orbital data centers. The satellites would rely primarily on optical inter-satellite links for data transfer, using Blue Origin’s TeraWave network as their ground backhaul infrastructure. It represents Blue Origin’s entry into the space-based computing market being pursued simultaneously by SpaceX and Starcloud.

How many satellites does Project Sunrise involve?

The FCC application authorizes Blue Origin to operate up to 51,600 satellites. They would operate in circular sun-synchronous orbits between 500 and 1,800 kilometers altitude, with orbital inclinations ranging from 97 to 104 degrees, and each orbital plane containing approximately 300 to 1,000 satellites.

Why are sun-synchronous orbits chosen for orbital data centers?

Satellites in dawn-dusk sun-synchronous orbits receive near-continuous sunlight, enabling steady solar power generation without the interruptions that affect other orbital regimes. This eliminates the need to draw from terrestrial electrical grids and removes constraints around energy availability that limit data center expansion on the ground, making them practical candidates for power-hungry computing workloads.

How does Project Sunrise differ from Blue Origin’s TeraWave constellation?

TeraWave is a 5,408-satellite communications network designed to deliver enterprise connectivity at up to six terabits per second. Project Sunrise uses satellites as actual computing infrastructure – orbital data centers – rather than as data transmission conduits. The two systems are designed to work together, with TeraWave providing the backhaul connectivity that Project Sunrise satellites use to move compute results to and from ground-based clients.

What spectrum does Project Sunrise use?

Project Sunrise will use the Ka-band exclusively for telemetry, tracking, and command functions, specifically 18.8 to 19.3 GHz for space-to-Earth transmissions and 28.6 to 29.1 GHz for Earth-to-space. These operations are conducted on a non-interference, non-protected basis, meaning licensed primary users take precedence. Primary data traffic moves through optical inter-satellite links, which require no radio frequency spectrum allocation.

What regulatory waivers is Blue Origin seeking for Project Sunrise?

Blue Origin is requesting four primary waivers: exemption from FCC processing round procedures, exemption from milestone and surety bond requirements, a waiver of the filing completeness rule (because the satellite design is still maturing), and a waiver of Schedule S data requirements (because submitting complete orbital parameters for 51,600 satellites in the web form is impractical). Each waiver is grounded in existing FCC precedents and the agency’s own pending Space Modernization rulemaking.

Who are the main competitors to Project Sunrise?

SpaceX has filed for up to one million orbital data center satellites using the Starlink infrastructure as backhaul. Starcloud, an Nvidia-backed Redmond-based startup, filed for 88,000 satellites in February 2026 and has already deployed hardware in orbit. China has announced a 200,000-satellite constellation focused on state-coordinated in-orbit processing. All proposals target the same fundamental problem: terrestrial data center capacity running out of room to expand.

What are the main technical challenges facing orbital data centers?

Radiation hardening for high-performance processors, thermal management without atmospheric cooling, debris mitigation across densely populated orbital bands, and the engineering complexity of manufacturing thousands of satellites at competitive cost are the primary challenges. No company has yet demonstrated a commercial-scale orbital data center, and the cost-competitiveness threshold identified in Google’s November 2025 feasibility study isn’t expected to arrive until approximately 2035 under optimistic assumptions.

What role does New Glenn play in Project Sunrise?

New Glenn, Blue Origin’s heavy-lift rocket that first reached orbit on January 16, 2025, is the intended launch vehicle for both Project Sunrise and TeraWave. The company is also developing the New Glenn 9×4 super-heavy variant, which can lift more than 70,000 kilograms to low Earth orbit – more than 50 percent more than the current design. Having its own launch vehicle is a significant advantage that reduces Blue Origin’s dependence on external providers and avoids the launch capacity bottleneck currently affecting Amazon Leo.

What is the broader significance of the orbital data center race?

The race to build orbital data centers reflects a convergence of two major technological pressures: AI compute demand that is straining terrestrial infrastructure, and falling launch costs that are making large-scale constellation deployment economically plausible. The outcome will likely reshape both the data center industry and the competitive structure of the space sector, with companies that control both launch capability and orbital infrastructure – Blue Origin and SpaceX being the primary examples – holding structural advantages that pure satellite operators can’t easily replicate.