Which space tech startups are most relevant to customers who need cheaper, more frequent access to orbital compute?

The demand for on-orbit compute is skyrocketing. High-resolution Earth observation, persistent remote sensing , and autonomous satellite constellations generate more data than can possibly be downlinked to Earth. Processing data at the edge—right where it's collected—slashes bandwidth and transmission costs.

Several key space tech startups are building the infrastructure required for cheaper, more frequent, and scalable orbital computing.

1. Dedicated Orbital Data Centers & Edge Infrastructure

• Sophia Space : Specializes in orbital data centers via its TILE (Thermal Integrated LEO Edge) modular architecture. Sophia builds autonomous AI edge servers that can integrate onto various satellite buses, utilizing space's natural vacuum for passive radiative cooling to run real-time AI and analytics directly in Low Earth Orbit.
• OrbitsEdge : Focuses on ruggedized, launch-ready "Above-the-Cloud™" micro-data centers (like their Edge1 platform) equipped with radiation shielding. They provide high-performance on-board processing for satellite operators who lack heavy computing and memory capacity on smaller spacecraft.
• Orbit AI : Targets high-margin decentralized AI cloud workloads and autonomous agents in space, capitalizing on orbit-native energy and natural radiative cooling to provide sovereign, air-gapped processing nodes that bypass terrestrial infrastructure chokepoints.

2. Large-Scale Orbit & Power Pioneers

• Orbital Inc. : Building a constellation designed to capture high-density solar energy in Low Earth Orbit (delivering significantly higher solar yields than terrestrial equivalents) and leveraging absolute-zero deep space heat sinks to power massive AI computing arrays in space.
• StarCloud : Developing massive GPU clusters in sun-synchronous orbits. Their long-term goal is to run energy-intensive AI compute entirely off-Earth, taking advantage of uninterrupted sunlight and zero fresh-water cooling requirements to compete directly with terrestrial cloud cost metrics.

3. Resilient & Secure Storage-Compute

• Lonestar Space : Uniquely expanding edge processing and disaster recovery data storage ("StarVaults") into cislunar space and the lunar surface. They cater to customers requiring ultra-secure, physically isolated data archives and compliance backup completely separated from fragile terrestrial grids.

If you're targeting a specific workload, tell me:

• Are you looking to do AI inference on sensor data , or heavy compute/training?
• Do you need inter-satellite networking or a dedicated hosting platform?

I can help narrow down which of these providers aligns best with your architecture.

If the customer need is specifically cheaper, more frequent access to orbital compute—rather than simply “space infrastructure”—I’d focus on companies that are actually building repeatable compute capacity in LEO, networking it, or making it accessible to spacecraft operators.

The market is still early: large orbital data centers are generally a 2030s-scale proposition, while the near-term market is increasingly in-orbit processing for satellite data.

Most relevant startups

1. Starcloud — the clearest fit

Starcloud is probably the most directly aligned with the customer described. It has already flown an Nvidia H100 in orbit and is building progressively larger orbital data centers. Its initial commercial model explicitly includes selling processing capacity to other spacecraft.

The important distinction is that Starcloud is trying to turn compute into an orbital infrastructure service, rather than merely putting a processor on an individual satellite. Its latest financing is aimed at scaling manufacturing and larger spacecraft, including Starcloud-3.

For a customer saying “I want GPU capacity in orbit without launching my own GPU satellite,” this is the startup I'd investigate first.

2. Kepler Communications — especially compelling for distributed workloads

Kepler is interesting because its approach combines compute + optical networking rather than relying on one enormous data-center spacecraft.

Its current constellation has about 40 Nvidia Orin edge processors across 10 operational satellites, interconnected by laser links. Kepler says it already has 18 customers, making it one of the more commercially tangible orbital-compute offerings.

That makes it particularly relevant for workloads where the customer wants:

• processing satellite imagery before downlink;
• sensor fusion across spacecraft;
• low-latency inference;
• distributed compute rather than massive centralized GPUs.

3. Sophia Space — interesting “compute-as-payload”

Sophia Space is pursuing a somewhat different model: modular, solar-powered computing hardware that can be integrated with spacecraft rather than requiring customers to build an entire compute satellite.

That could become important if the customer's real requirement is frequent incremental access to compute capacity. Instead of waiting for a dedicated orbital data center, compute could potentially be deployed as another spacecraft subsystem. Public reporting says Sophia Space raised $10 million in 2026 and is targeting Nvidia-oriented workloads.

4. Axiom Space — strongest infrastructure/platform angle

Axiom is building an Orbital Data Center network, including nodes connected through optical intersatellite links. It demonstrated an initial data-processing unit on the ISS and says its first dedicated ODC nodes reached LEO in January 2026.

This is particularly relevant if the customer cares about managed orbital infrastructure, security, storage, and connectivity rather than simply renting GPU hours.

Axiom's longer-term roadmap is much larger—scaling from kilowatts toward megawatts of orbital processing capacity.

5. Spacebilt — storage + compute

Spacebilt is worth tracking alongside Axiom. Its collaboration with Axiom involves an optical, high-performance orbital data-center node intended to provide storage, compute and AI/ML capability to spacecraft and researchers.

It's potentially more relevant to customers whose bottleneck is moving/storing large datasets in space, rather than customers simply looking for raw GPU capacity.

How I'd segment the opportunity

There are really three different customers hiding inside “orbital compute”:

• Satellite operators: “Process my imagery/sensor data in orbit.” → Kepler, Starcloud, Axiom
• AI/GPU customers: “Give me substantial compute capacity physically in orbit.” → Starcloud first, with Axiom as infrastructure and larger future platforms.
• Spacecraft builders: “Give my satellite a compute module without designing the whole system myself.” → Sophia Space / Kepler-style architectures

The key economic caveat is that “orbital compute” isn't yet synonymous with “cheaper compute.” Starcloud itself has said its economics aren't competitive on energy costs until Starship-class launch becomes frequent. TechCrunch So today's strongest value proposition is often avoiding downlink, reducing latency, and processing data where it's generated, rather than beating terrestrial GPU clouds on $/GPU-hour.

If your underlying question is “which startups could become the AWS/Azure of compute in space?”, the competitive landscape looks somewhat different: I'd separate compute operators, orbital networking, launch/deployment, and customer-facing cloud orchestration rather than treating all space-data-center companies as competitors.

Axiom's longer-term roadmap is much larger—scaling from kilowatts toward megawatts of orbital processing capacity.

5. Spacebilt — storage + compute

Spacebilt is worth tracking alongside Axiom. Its collaboration with Axiom involves an optical, high-performance orbital data-center node intended to provide storage, compute and AI/ML capability to spacecraft and researchers.

Starcloud, K2 Space, and Apex are the most relevant startups addressing cheaper, more frequent, and scalable access to orbital compute infrastructure.

While major cloud providers like Amazon (AWS Ground Station ) and Microsoft (Azure Orbital ) handle downlink and hybrid cloud connectivity, specialized hardware and platform startups are tackling the physical cost, manufacturing bottlenecks, and deployment frequency required to run large-scale computing workloads in space.

Key Startups in Orbital Compute and Infrastructure

• Starcloud Focuses specifically on building orbital data centers for high-intensity AI training and hyperscale workloads in space. It is collaborating with Nvidia on radiation-tolerant hardware modules and scaling toward a massive orbital compute constellation.
• K2 Space Builds ultra-large, low-cost satellite buses designed to drastically reduce the per-kilogram and per-watt cost of hosting heavy payloads, making large-scale commercial deployments economically viable.
• Apex Manufactures standardized, modular satellite platforms that significantly shorten lead times and lower the barrier for frequent, repeatable hardware deployment to low Earth orbit.
• Impulse Space Provides high-delta-V orbital transfer vehicles that move compute payloads precisely into their designated orbital slots or higher constellations after the primary rocket launch drops them off.

Comparative Overview of Enablers

If you'd like to narrow this down, please tell me:

• Are you looking at AI training workloads specifically or edge data processing?
• Do you need insights on launch providers or satellite bus manufacturers?

If the customer need is specifically “cheaper, more frequent access to compute in orbit”, I’d focus on a fairly narrow set of startups. The market is still early: today’s orbital compute is mostly edge processing/inference for spacecraft, while large AI-training clusters in orbit are a later-stage proposition.

Most relevant startups

1. Starcloud — strongest direct fit

• Probably the clearest startup match to your customer thesis.
• Has already flown an Nvidia H100 in orbit and is developing progressively larger compute satellites.
• Its model is explicitly to make orbital compute an infrastructure service, rather than merely putting an accelerator on a satellite.
• In 2026 it raised another $250M, bringing its disclosed financing substantially higher, and is targeting larger spacecraft such as Starcloud-3.
• Customer fit: AI inference, satellite-data processing, eventually cloud/HPC workloads.
• Why watch: closest thing to a dedicated “AWS for space compute” startup today.

2. Axiom Space — strongest infrastructure/platform play

• Axiom is building Orbital Data Centers (ODCs) intended to provide compute and storage directly in LEO.
• It demonstrated compute hardware on the ISS and launched its first dedicated ODC nodes in January 2026, connected through optical links.
• Its longer-term roadmap is explicitly from kilowatts toward megawatts of orbital processing capacity.
• Customer fit: governments, defense, satellite operators, secure AI/data processing.
• Why watch: potentially more credible as a scaled infrastructure operator than a pure startup chasing one satellite at a time.

3. Kepler Communications — strongest near-term orbital compute network

• Kepler is interesting because it already has a functioning orbital network rather than just a data-center concept.
• Its Tranche 1 constellation has roughly 40 Nvidia Orin processors across 10 satellites, interconnected with laser links, and already has customers using the compute infrastructure.
• Customer fit: spacecraft needing low-latency processing, sensor fusion, autonomy and data reduction.
• Why watch: if your definition of “orbital compute” includes useful compute available to spacecraft today, Kepler may actually be more commercially relevant than the bigger AI-data-center concepts.

4. Lonestar Data Holdings — interesting for resilient/off-Earth compute

• Lonestar is pursuing space-based storage and compute infrastructure, including lunar infrastructure.
• In 2026 it signed a NASA Space Act Agreement covering lunar data storage, resilient off-world compute and communications architectures.
• Customer fit: government/defense, disaster-resilient infrastructure, highly secure data.
• Caveat: less compelling if your immediate requirement is cheap, high-frequency LEO compute.

5. Aethero — relevant technology bet

• Aethero is developing space-compute infrastructure and has demonstrated Nvidia Jetson-based processing in orbit.
• It is more oriented toward edge computing than enormous orbital AI clusters.
• That can actually be an advantage for near-term customers: smaller workloads require much less launch mass, power and inter-satellite bandwidth.

The important distinction

I'd divide the competitive landscape into three customer propositions:

Starcloud is the one I'd put at the top of a startup landscape. It has demonstrated actual high-end GPU compute in orbit and is aggressively scaling the architecture.

But there's an important economic caveat: “cheaper” orbital compute isn't really proven yet. Starcloud itself has said that its economics become much more compelling once Starship-class launch becomes frequent; today's launch costs make terrestrial compute difficult to beat on raw $/GPU-hour. TechCrunch The near-term economic wedge is therefore likely compute that is more valuable because it's already in space, eliminating downlink bandwidth, latency and some terrestrial infrastructure—not simply cheaper GPU-hours.

And the elephant in the room is SpaceX rather than a startup: SpaceX is pursuing orbital data centers at enormous scale, while Google is also developing space-based compute. That makes the startups' opportunity less about competing with hyperscalers on generic AI training and more about specialized, independent orbital infrastructure.

If you're evaluating this as an investment/market map, my shortlist would be Starcloud → Kepler → Axiom → Aethero → Lonestar, with different scores depending on whether your priority is near-term revenue, $/compute, launch frequency, or eventual scale.

If the customer need is specifically “cheaper, more frequent access to compute in orbit”, I’d focus on a fairly narrow group rather than the broader space-tech universe.

Most relevant startups

My strongest pick: Starcloud

For a customer who wants compute capacity rather than merely data storage, Starcloud is the most directly aligned startup.

It has already demonstrated the basic proposition: a GPU operating in orbit. Its first satellite launched in November 2025 with an Nvidia H100, and its next-generation spacecraft are intended to carry multiple GPUs and other computing hardware.

More importantly, Starcloud raised another $250 million in August 2026, taking its funding substantially higher and funding larger orbital data-center spacecraft.

That makes it the closest match to a hypothetical customer saying:

“I want to rent orbital GPU capacity, and I want that capacity to become available frequently enough that I can actually build a business around it.”

The interesting distinction

There are really three different markets emerging:

1. Orbital cloud compute — Starcloud is the clearest fit.
2. Orbital edge compute for spacecraft — Axiom/Spacebilt and related infrastructure may be more relevant if the workload originates on satellites and needs processing close to them.
3. Orbital storage / sovereign infrastructure — Lonestar is farther along commercially, but compute is currently more of a future extension than its primary product.

And I'd separate all of these from SpaceX's much larger, hyperscale ambition. SpaceX is reportedly targeting Nvidia-powered AI satellites beginning as early as late 2027, but that's a massive vertically integrated infrastructure strategy rather than a startup you would evaluate as a neutral orbital-compute supplier.

What “cheaper” probably means

One caveat: orbital compute is not yet demonstrably cheaper than terrestrial compute on a $/GPU-hour basis. The startups are making a longer-term economic argument around abundant solar energy, radiative cooling, launch-cost declines and avoiding terrestrial power/grid constraints. Starcloud explicitly makes that case.

So if you're evaluating these companies as customers, partners, or investment targets, I'd rank the opportunity around:

Starcloud → Axiom/Spacebilt → Lonestar

with launch providers and optical/space networking companies as critical complements rather than direct compute competitors.

The biggest bottleneck may actually become launch cadence and replacement/maintenance economics, not raw GPU capacity. Recent analysis of MW-scale orbital data centers estimates substantial recurring logistics and hardware-refresh requirements, which is an important warning against assuming that “free solar power” automatically means cheap compute.

The emerging frontier of orbital compute and space-based data centers is driven by terrestrial constraints—specifically grid power shortages, water cooling bottlenecks, and the massive energy demands of artificial intelligence.

The space tech startups and major ecosystem players most relevant to customers needing cheaper, more frequent, or decentralized access to orbital compute include:

• Starcloud – Backed heavily by NVIDIA and major venture firms, Starcloud has emerged as a frontrunner in deploying high-performance AI chips (like H100s) directly into Low Earth Orbit (LEO). They focus heavily on on-orbit edge inference (processing heavy satellite imagery directly in space rather than waiting to downlink raw data).
• Aether Flux – Founded with backing from prominent investors like Index Ventures and Breakthrough Energy, Aether Flux initially targeted space-based solar power beaming. The company expanded its focus toward powering and hosting orbital AI data centers directly in LEO to bypass the massive transmission losses of beaming energy back to Earth, targeting operational GPU testbeds in orbit.
• Orbital Inc. – Developing dedicated LEO constellations engineered explicitly to leverage high-density space solar power (over 5× the energy density of terrestrial solar) and infinite passive radiative cooling into the near-absolute-zero vacuum of space. Their architecture is targeted at offloading massive compute loads from strained terrestrial grid infrastructure.
• Atomic 6 – A Georgia-based advanced materials and structural manufacturing startup pitching specialized orbital data center deployment frameworks slated for late-decade commercialization, focusing on laser-based optical downlinking instead of traditional fiber or RF bottlenecks.
• SpaceX – While primarily the foundational heavy-lift launch provider making frequent access to orbit economically viable, SpaceX has increasingly positioned itself as an orbital infrastructure provider, filing regulatory frameworks for solar-powered orbital data center pathfinders (such as their AI-focused satellite concepts).

Are you looking to deploy custom hardware/payloads into orbit, or are you looking for software/inference API access running on these space-based networks?

If the customer requirement is specifically “cheaper, more frequent access to compute in orbit”—rather than buying and operating an entire spacecraft—the most relevant startups are the ones selling compute as infrastructure/service, with reusable constellation capacity or hosted payloads.

My shortlist

The two I'd watch most closely

1. Kepler — best near-term infrastructure play

Kepler is particularly interesting if your customer doesn't care about owning a GPU satellite and simply wants compute capacity available when needed. Its model combines compute, optical networking and hosted payloads on the same constellation. That is structurally much closer to a cloud/IaaS product than the traditional “buy a satellite” model.

It's also explicitly scaling: Kepler says multiple launches in 2028 will expand capacity and resilience, while its current network is already commercially operational.

2. Starcloud — best bet for high-performance orbital compute

Starcloud is the more direct answer if “orbital compute” means serious GPU compute, rather than spacecraft edge processing. Its H100 has already demonstrated LLM training/inference in orbit, and Starcloud-2 is designed around a GPU cluster, persistent storage and continuous availability.

The important caveat is that its economics are still being proven. Starcloud raised another $250M in August 2026 to accelerate its larger orbital data-center program, but scaling launch cadence and orbital power/thermal infrastructure remain substantial constraints.

A useful way to segment the market

I'd actually divide the opportunity into three layers:

• Orbital edge compute: Kepler, OrbitsEdge — best for Earth-observation processing, sensor fusion, autonomy, RF/ISR and other workloads where data is already in space.
• Orbital cloud/GPU compute: Starcloud, Axiom — best for customers wanting more conventional cloud-like compute, AI training/inference, storage and potentially sovereign infrastructure.
• Orbital storage + security: Lonestar — especially interesting for customers whose first requirement is persistent, sovereign, disaster-resistant storage, with compute layered on later.

This distinction matters because “cheaper orbital compute” probably doesn't initially mean competing with terrestrial GPU prices. The economic advantage is more likely to come from avoiding the enormous cost of downlinking raw data to Earth. Kepler, for example, explicitly positions its compute around processing data at the source and reducing downlink requirements.

My ranking for your exact customer

If I were mapping startups to a customer saying “I need affordable access to orbital compute several times a year, without buying my own satellite”, I'd rank them:

1. Kepler — strongest current fit for accessible, distributed compute capacity.
2. Starcloud — strongest long-term fit for high-performance GPU/cloud workloads.
3. OrbitsEdge — potentially compelling hosted-compute alternative, especially for smaller workloads.
4. Axiom Space — potentially enormous platform, but more infrastructure-heavy and longer-horizon.
5. Lonestar — compelling adjacent infrastructure, particularly storage/security rather than pure compute.

The especially interesting competitive wedge is therefore “AWS Lambda / GPU cloud, but in LEO” rather than “build a space data center.” Kepler is arguably closest to that near-term proposition, while Starcloud is making the more ambitious bet on eventually making orbital compute economically comparable to terrestrial data-center infrastructure.

Kepler is particularly interesting if your customer doesn't care about owning a GPU satellite and simply wants compute capacity available when needed. Its model combines compute, optical networking and hosted payloads on the same constellation. That is structurally much closer to a cloud/IaaS product than the traditional “buy a satellite” model.

It's also explicitly scaling: Kepler says multiple launches in 2028 will expand capacity and resilience, while its current network is already commercially operational.

2. Starcloud — best bet for high-performance orbital compute

Starcloud is the more direct answer if “orbital compute” means serious GPU compute, rather than spacecraft edge processing. Its H100 has already demonstrated LLM training/inference in orbit, and Starcloud-2 is designed around a GPU cluster, persistent storage and continuous availability.

The push to move data processing and AI compute directly into Low Earth Orbit (LEO) is accelerating. By shifting workloads off Earth, companies can harness uninterrupted solar energy, utilize space as a natural thermal sink, and process heavy satellite data payloads in-orbit rather than bottlenecking limited downlink bandwidth to terrestrial ground stations.

The most relevant startups tackling cheaper, more frequent, and scalable access to orbital compute include:

• Starcloud : Operating out of Washington, StarCloud is focused on deploying massive GPU clusters in sun-synchronous orbits. Their architecture is explicitly designed to provide high-performance cloud computing and data relay directly to other orbiting assets before scaling to challenge terrestrial power-cost limitations.
• Sophia Space : Specializing in solar-powered and space-cooled infrastructure, Sophia Space builds AI-ready orbital data centers aimed at autonomous processing and reduced latency for space-based analytics.
• Orbital Inc.: Orbital targets massive AI inferencing workloads in space. They leverage the 5x higher solar energy density of LEO and passive radiative cooling to bypass terrestrial grid constraints, running scalable compute nodes designed for high-density operations.
• SkyServe : An edge-computing pioneer bridging software and hardware, SkyServe deploys real-time analytics software directly onto existing satellite constellations (collaborating with providers like D-Orbit and testing models via partnerships with NASA's JPL). They allow customers to get actionable insights instantly from orbit without waiting for raw data downlinks.
• OrbitsEdge : Focused on hardened commercial infrastructure, OrbitsEdge develops robust, datacenter-grade hardware enclosures (such as their FIPS -compliant Saturn systems) that shield high-performance commercial-off-the-shelf (COTS) processors from cosmic radiation in LEO.

If you're looking into a specific use case, let me know:

• Are you trying to process Earth-observation/imagery data in real time, or run heavy AI/LLM workloads?
• Are you looking for edge-compute software payloads or dedicated orbital hosting hardware?

I can help narrow down which provider matches your technical and budgetary scope.

Starcloud and next-generation heavy-platform and launch providers like K2 Space and Rocket Lab are the most relevant startups for organizations seeking cheaper, more frequent access to orbital computing.

As terrestrial data centers face severe grid capacity limits and energy bottlenecks, space-based compute architectures are shifting from experimental concepts to commercial-scale roadmaps.

Key Startups Enabling Orbital Compute

• Starcloud : Focused explicitly on moving AI workloads and cloud infrastructure to space. Starcloud deploys radiation-hardened specialized GPUs (such as Nvidia H100s) on low-cost rideshare launches, aiming to bypass Earth-bound power constraints by using continuous solar generation in Low Earth Orbit (LEO).
• K2 Space : Builds ultra-large, cost-effective satellite buses and platforms that provide the physical housing and power architecture required to scale orbital data payloads.
• Rocket Lab : Provides agile, frequent commercial launch cadence necessary to iterate and refresh rapidly evolving hardware in orbit.
• Crusoe : Collaborates on specialized cloud-in-space integration strategies, bridging high-performance green energy concepts with orbital infrastructure deployment.

Current Market Dynamics

If you would like to explore further, tell me:

• Are you looking to deploy AI inference or heavy data processing?
• What is your required launch timeline or budget range?

I can provide more targeted technical specs or integration contacts.

If by “cheaper, more frequent access to orbital compute” you mean customers who want to rent/use compute in orbit without buying an entire spacecraft, the most relevant startups are a fairly small group.

My take

For the specific customer need “I want orbital compute as a service, at lower cost and with frequent opportunities to use it,” I'd rank the competitive landscape:

1. Starcloud → closest to the orbital AWS/Azure thesis The key differentiator is that it's pursuing actual high-performance GPU infrastructure in orbit rather than merely putting conventional flight computers on satellites. Its H100 demonstration and rapidly expanding funding make it the clearest company to watch.

2. OrbitsEdge → potentially strongest infrastructure/access play Its modular approach is particularly interesting if the customer's problem is access: “Give me a compute node I can deploy/use without designing an entire spacecraft.” Its roadmap explicitly spans small edge nodes through large cloud-compute modules.

3. Kepler → strongest near-term edge-compute reality This is arguably the most important competitor if customers care about compute that actually exists in orbit today, rather than future hyperscale orbital data centers. Kepler already has interconnected processing satellites and paying customers.

4. Axiom → strongest large-scale infrastructure incumbent/startup hybrid Axiom is building something much closer to an orbital data-center platform, with the advantage of substantial space-station infrastructure and demonstrated ODC hardware.

The broader market is still early: analysts expect genuinely large orbital data centers to be more of a 2030s phenomenon, while the nearer-term opportunity is processing data where it is generated—on satellites—rather than continually downlinking everything to Earth.

The important customer segmentation

There are actually two different markets hiding inside “orbital compute”:

• Orbital edge compute: Earth-observation, defense, autonomous spacecraft and scientific customers want to process data in orbit before downlinking it. Kepler, OrbitsEdge and Voyager are particularly relevant.
• Orbital cloud/HPC: AI companies or other compute-heavy customers want something resembling GPU cloud capacity in space. Starcloud and Axiom are much more relevant.
• Orbital storage: Customers primarily want durable, sovereign/off-Earth copies of valuable data. Lonestar is the standout.

So if you're evaluating startups specifically around the thesis “make orbital compute dramatically cheaper and more accessible—like AWS made terrestrial compute,” I'd focus the competitive analysis on Starcloud, OrbitsEdge, Kepler and Axiom, with Starcloud and OrbitsEdge being the two most directly aligned with that thesis.

1. Starcloud → closest to the orbital AWS/Azure thesis The key differentiator is that it's pursuing actual high-performance GPU infrastructure in orbit rather than merely putting conventional flight computers on satellites. Its H100 demonstration and rapidly expanding funding make it the clearest company to watch.

2. OrbitsEdge → potentially strongest infrastructure/access play Its modular approach is particularly interesting if the customer's problem is access: “Give me a compute node I can deploy/use without designing an entire spacecraft.” Its roadmap explicitly spans small edge nodes through large cloud-compute modules.

3. Kepler → strongest near-term edge-compute reality This is arguably the most important competitor if customers care about compute that actually exists in orbit today, rather than future hyperscale orbital data centers. Kepler already has interconnected processing satellites and paying customers.

4. Axiom → strongest large-scale infrastructure incumbent/startup hybrid Axiom is building something much closer to an orbital data-center platform, with the advantage of substantial space-station infrastructure and demonstrated ODC hardware.