Intro: Every week, another announcement promises to put a data center in space. But what these announcements rarely acknowledge is that the foundational infrastructure of orbital data centers (ODCs) – onboard compute, storage, and connectivity – is being built and validated in orbit today. This isn’t some Big Bang moment waiting to happen, nor will it arrive as a single breakthrough. ODCs are being built the way critical infrastructure has often been built: incrementally, mission by mission, with each step proving the next.
The market misread
The market may not be fully registering this moment in three ways. The first is structural: orbital compute is very often being treated as a future roadmap item, rather than a deployable infrastructure layer already serving real use cases today. In our view, the architectural shift that makes ODCs possible – from custom, one-off hardware to a programmable, software-defined platform – has already happened. That shift means certain workloads already make economic sense in orbit today: latency-tolerant AI inference, Earth-observation processing at the point of collection, batch analytics, and sovereign workloads that cannot sustain certain operational ground loops. Beyond economics, some services are inherently space-native: autonomous satellite operations, in-orbit coordination, and space-domain awareness don’t simply benefit from orbital compute; they require it to scale.
Secondly, there’s still a prevailing assumption that space-grade hardware means slow and expensive, and that radiation hardening comes at the cost of performance. In our experience, that’s no longer the case today. Closing that gap, however, doesn’t happen via using component as-is per datasheet; it requires flight heritage, a rigorous understanding of the space radiation environment, and design methodologies that can only be built through years of operational experience. This is what “radiation-resilient-by-design” actually means in practice: applying those hard-won principles to advanced commercial technologies to maximise performance and SWaP, without compromising mission reliability and cost. This is not a problem you can solve through theoretical specifications alone— but it is one that companies with genuine operational experience are actively solving.
The third assumption worth addressing is displacement: that orbital compute is all about replacing ground infrastructure. This likely misses the point. The ODC adds a new layer to the equation, shifting the centre of gravity between space and ground processing gradually and for specific mission classes. This is not a displacement story and it won’t happen overnight; it’s an evolution of the overall compute architecture, and one that makes both layers more valuable, not less.
The financial case for on orbit compute
A ground-based processing model assumes you can get the data down, but for a growing class of missions, that assumption is breaking. One clear financial lever is now the value of compressing the time between detection and decision. For commercial EO customers, that’s the difference between a near-real-time analytics product commanding a premium and a daily batch delivery that’s become a commodity. For government customers, real-time processing isn’t necessarily a premium feature but can be a new use case entirely. Threat detection, situational awareness, time-critical response: these require very rapid decisions and that creates capability streams that are simply harder to achieve in a ground-processing model.
The sovereign dimension adds a constraint that makes the financial case for orbital compute capabilities even sharper: many government customers can only route data through ground stations on sovereign territory. That ground loop is not always operationally possible, and sometimes not legally permissible. Onboard processing can significantly reduce or remove the dependency entirely.
Finally, resource scarcity on the ground is a constraint that is increasingly well understood by the industry. Power grid limitations, cooling costs, and real-estate availability in certain markets are forcing operators to rethink where processing happens. Orbital compute is increasingly part of that conversation, not as a default answer, but as a genuinely practical option for the right workloads.
As technology matures and economics improve, the financial case, when properly framed, may no longer be “orbital compute vs ground infrastructure “it’s” what decisions can you make, and what revenue can you generate, that are difficult or impossible to achieve today”.
The real path to space data centers
Returning to the structural question, it’s important to define what an ODC is, and what it should look like, given that the market narrative sometimes treats it as a breakthrough still waiting to happen. The way we see it, a genuine ODC is not a satellite with onboard processing alone, it’s a shared, accessible compute infrastructure in orbit that any authorised user or mission can reach, much like a ground data center serves multiple tenants across multiple workloads.
That requires: programmable, multi-tenant compute capable of running arbitrary workloads, inference, signal processing, sensor fusion; persistent storage sufficient to buffer mission data across contact windows; high-bandwidth downlink that doesn’t negate the onboard processing advantage; and ground-side orchestration that treats the orbital node as part of a hybrid architecture, not an isolated edge device.
Standards are likely the enabler that makes shared infrastructure real: just as ground data centers became commercially scalable when access, interfaces, and workload portability were standardised, an ODC will likely require equivalent standardisation so that any operator can connect, deploy and consume without bespoke integration for every mission
Of this stack, the compute and storage layers are arguably the most mature today. Connectivity is currently the ceiling. Orchestration and standards are the frontier and where some of the most important work is now happening.
Ramon.Space’s current portfolio is designed to addresses this stack today: a software-defined platform with custom interfaces, advanced processors and FPGAs including AI engines, and flight proven high-capacity TB storage, integrated into a single onboard processor unit.
Our next generation NuCore platform advances every dimension of that stack processing, storage and connectivity, and represents our most ambitious answer to what a near-term space data center node will require.
The seat at the table
Our best advice is don’t wait for the perfect full-stack ODC. The technology that exists today can already address real operational problems and generate real value. Bring your business model and your data problem to the conversation, and set aside preconceived ideas about space being too slow or too expensive, the economics and the capabilities may have moved faster than the perception.
Engage early with companies that have genuine heritage and deep technical capability. The knowledge gap between those who have built these systems and those who haven’t is wider than it appears from the outside.
Lastly, in our view the most credible path to scale is not a single massive deployment. It is a sequence of real missions, each one building on the last and as that infrastructure scales, so will the ecosystem and application layer around it. The companies and operators who understand that are already moving. The question for everyone else is whether they engage now or find themselves waiting for an announcement that tells them the moment has passed.