Every major risk model a central bank, sovereign wealth fund or clearing house runs is subject to the legal and physical boundaries of the data centre it runs in. Regulators can compel disclosure, adversaries can compel shutdown, and physical infrastructure can be seized or sanctioned. Moving compute off-planet places the execution environment in a jurisdiction that does not yet exist in international law — a deliberate ambiguity that some financial actors will find operationally attractive and others will find existentially necessary.
The satellite stack here is not about connectivity; it is the computer. A constellation of compute microsatellites, each carrying a radiation-tolerant GPU or FPGA array equivalent to several petaFLOPS of sustained throughput, executes encrypted financial workloads uplinked in ciphertext. Results are downlinked only to credentialed ground terminals. On-board homomorphic or secure-enclave processing means the satellite operator — even the launching nation — cannot read the plaintext of the computation in flight. The orbital environment also provides a natural physical air-gap: no fibre tap, no server-room entry, no emergency court order reaches low Earth orbit within the execution window of a millisecond-scale model run.
Operationally, the first viable use case is not exotic: it is disaster-continuity compute for a sovereign central bank that needs its settlement and risk systems to keep running if every domestic data centre is taken offline by conflict, cyberattack or natural catastrophe. The second use case, longer-horizon, is competitive advantage — running proprietary macro models and portfolio optimisations in an environment where no foreign intelligence service or competitor can observe the workload pattern, timing or result. Both cases reward a nation that owns the constellation outright rather than queuing behind a commercial provider whose government has its own interests.