Flower Labs and Starcloud Reach a Major AI Milestone in Orbit
AI infrastructure is expanding beyond Earth, with space emerging as an increasingly important environment for how and where AI systems operate. As decentralized AI systems built using Flower bring compute and learning closer to where data is generated, the assumptions that shaped centralized, cloud-based machine learning begin to break down.
Today, Flower Labs and Starcloud are sharing a major milestone in this shift: the successful execution of a decentralized AI workload using Flower on an operational Starcloud satellite.
This milestone shows that powerful, adaptive, end-to-end AI workloads — once limited to earth-based infrastructure, can now be deployed, updated, and validated directly in space. Just as importantly, it highlights the role of the Flower decentralized AI frameworks in enabling this transition, and how rapidly the space-based AI domain is advancing when paired with Starcloud’s satellite data center infrastructure.
Decentralized AI in Orbit
The mission involved locally fine-tuning a Vision Transformer (ViT) model directly on a Starcloud satellite. The model was adapted in orbit to classify satellite imagery, including urban areas, forests, lakes, and other land-cover features, directly on-board the satellite — where the data is generated.
Running this AI workload in space requires more than simply deploying a model to a new environment. Space systems operate under intermittent connectivity, constrained bandwidth, and limited opportunities for direct intervention. Centralized training pipelines, which depend on continuous access to large data pools and persistent cloud connectivity, are poorly suited to these conditions.
Flower is the world’s most popular open-source decentralized AI framework and is designed for exactly this kind of environment. By enabling learning to happen across distributed nodes without requiring centralized data and compute, Flower makes it possible for models to be trained and updated where data lives; even when those nodes are in orbit.
In this mission, Flower enabled the orchestration of on-orbit fine-tuning while maintaining reliability and control despite the constraints of space-based systems. Telemetry retrieved from orbit confirmed that the workload executed as intended, validating that decentralized AI workflows can operate on real, operational satellites.
Why This Matters for AI Infrastructure
This milestone points to a broader change in how AI infrastructure is evolving.
As AI systems scale, centralized cloud data centers face increasing pressure from energy, cooling, latency, and bandwidth constraints. At the same time, vast amounts of valuable data are generated at the edge; by satellites, sensors, and distributed systems that are far removed from traditional compute hubs.
By enabling AI models to be updated and specialized directly in orbit, decentralized AI built using Flower allows intelligence to move closer to the source of data. This reduces dependence on downlink capacity, shortens the time from observation to insight, and enables satellites to deliver more relevant and actionable information.
Industry use cases already point in this direction. Space-based AI can support faster disaster response through early detection of wildfires or floods, improve maritime safety by identifying vessels or distress situations, enhance geospatial awareness for security and planning, and enable more autonomous satellite operations through real-time interpretation of telemetry and sensor data. All of these scenarios benefit from intelligence that can adapt at the edge rather than relying exclusively on centralized infrastructure on Earth.
A New Layer of the AI Stack
Orbital platforms also introduce new possibilities for AI infrastructure itself. Access to continuous solar energy, natural cooling in vacuum, and proximity to space-based sensors make orbit an increasingly attractive environment for certain classes of compute workloads.
Rather than replacing terrestrial data centers, space-based systems represent an extension of the AI stack: a complementary layer where decentralized intelligence can operate under different physical and operational constraints. Flower’s decentralized AI framework aligns naturally with this architecture, enabling learning to span heterogeneous, distributed environments across Earth and orbit. Starcloud’s in-space data center infrastructure provides the foundation that makes this new layer of the AI stack possible.
Looking Ahead
This is only the beginning.
This milestone shows that decentralized AI can run and adapt in space, pointing to a new class of distributed AI infrastructure.
Flower Labs looks forward to continuing our collaboration with Starcloud as we work toward larger, decentralized AI networks in orbit — and toward a future where intelligence is no longer limited by geography, connectivity, or gravity.