When we set out to design the software infrastructure for STELLAR's orbital network, we quickly realized that no existing operating system was designed for our requirements. Linux is designed for stable, connected hardware. Kubernetes is designed for data center workloads. FreeRTOS is designed for single embedded devices.
None of them are designed for a distributed system where: - Network partitions are regular and predictable (satellite passes) - Nodes have intermittent connectivity (ground station contacts) - Hardware operates in extreme environments (radiation, vacuum, temperature swings) - Workloads span quantum, classical GPU, and edge compute simultaneously - Latency to the operator is in the 10-100ms range at best
The Core Design Principles
Stellar OS is built on four principles that diverge from conventional operating system design:
1. Partition-First Design: Rather than treating network partitions as exceptional failures to recover from, Stellar OS treats them as the normal operating condition. Every service is designed to operate autonomously during pass gaps (the period when a satellite is not in contact with a ground station), then synchronize when connectivity is restored.
2. Heterogeneous Compute Abstraction: The OS presents a unified interface over quantum processors, classical GPUs, and neural inference accelerators. Workloads declare their compute requirements abstractly; the OS schedules them to appropriate hardware including migrating workloads between quantum and classical processors as coherence conditions change.
3. Radiation-Aware State Management: Radiation-induced bit flips are not bugs to be fixed they're environmental conditions to be tolerated. Stellar OS implements continuous state validation, automatic rollback to verified state snapshots, and workload migration when radiation events are detected.
4. Time-Sliced Orbital Resource Scheduling: Every orbital resource power, thermal, compute, communication bandwidth follows predictable patterns tied to orbital mechanics. Stellar OS schedules workloads around these patterns, ensuring peak compute workloads run when solar arrays are at maximum efficiency and thermal conditions are favorable.
The Quantum-Classical Interface
The most novel aspect of Stellar OS is the quantum-classical interface layer. Quantum processors are not general-purpose computers; they excel at specific problem types. The OS implements an automatic problem decomposition engine that identifies quantum-amenable subproblems within classical workloads and offloads them to the quantum processor.
This makes the quantum processor invisible to application developers they write classical code, and the OS accelerates it with quantum computation where beneficial.
Current Status
Stellar OS is in active development with our core team in Luxembourg and Abu Dhabi. We're targeting an initial SDK release in Q3 2026 that will allow research partners to begin developing orbital-native applications before our first satellite launches.
The operating system is, in many ways, the most important component of our stack. It determines what applications are possible on orbital infrastructure. We're building it to be the foundation for a century of orbital computing.