The systems architecture behind orbital data centers
Four systems that must work as one
Every STELLAR page has a visual reason to exist. The technology section separates compute, storage, thermal-power, and ground-cloud integration so each engineering discipline is understandable without turning the whole website into a document.
Orbital Compute
STELLAR compute nodes are designed for AI inference, autonomy workloads, data reduction, and high-value processing near orbital data sources.
Orbital Storage
Orbital storage gives missions a place to persist data, rank products, survive contact gaps, and deliver verified outputs to customers.
Thermal & Power
Compute in orbit only works when power generation, heat rejection, radiator area, duty cycles, and safe modes are designed as one system.
Ground + Cloud Integration
STELLAR connects orbital nodes to customer workflows through tasking, mission operations, ground stations, cloud handoff, and delivery verification.
Orbital compute is controlled by mission constraints
The website should show the technology like a product, not bury it in review language. Each system page uses a focused diagram and clear customer-facing explanation.
Orbital Compute payload — block schematic
Node-1 baseline · 500 km SSO · 95 kg allocation · pre-PDR
- J128 V regulated · 35 A
- J2SpaceWire ×2 · 200 Mbps
- J3CAN-FD · attitude / cmd
- J4Thermistor harness ×8
- J5TT&C bypass (S-band)
- P13.3 V · CPU + DDR · 6 A
- P212 V · FPGA core · 18 A
- T1Cold plate Cu ×4
- T2LHP loop ×2 (NH₃)
- T3Deployable radiator (0.42 m²)
- T4PT-1000 thermistor ×8
- T5Heater string ×4 (28 V)
- K1POST /jobs · submit
- K2GET /receipts
- K3GET /events
- K4POST /webhooks
- K5GET /telemetry
- K6GET /pass-schedule
- K7OAuth2 / mTLS guard
- K8Replay · audit log