For decades, the rule of satellite architecture was simple: keep the spacecraft dumb and the ground smart. Collect the data in orbit, downlink everything, and let racks of terrestrial compute do the thinking.
That rule is now being rewritten in public. SpaceX has announced plans for AI compute satellites built around next-generation Nvidia hardware. Alphabet's Project Suncatcher is testing whether its TPU accelerators can survive years of orbital radiation, with prototype spacecraft slated for 2027. Startups are manifesting orbital data-center demonstrators with each generation carrying dramatically more power and compute than the last. And across the established constellation operators, "edge AI" — running inference on the spacecraft itself, downlinking insights instead of terabytes of raw pixels — has quietly gone from experiment to operational practice.
Whether orbital data centers ultimately pencil out or not, the direction of travel is unmistakable: spacecraft are becoming computers, and computers are becoming spacecraft. Either way, the same thing gets scarce — the engineers who can build at that intersection.
Why onboard AI changes the engineering problem
Putting serious compute on a satellite touches nearly every subsystem discipline at once:
- Power and thermal — AI accelerators are power-hungry and hot. Solar array sizing, power electronics, and thermal rejection in vacuum become first-order design drivers, not afterthoughts.
- Radiation-tolerant electronics — commercial silicon meets the space environment. Engineers who understand SEU mitigation, redundancy architectures, and qualification of COTS compute are suddenly central to program success.
- FPGA and edge-inference design — much of today's operational onboard AI runs on FPGAs and specialized edge processors. RTL, verification, and DSP talent that also speaks ML is one of the rarest profiles in the industry.
- Flight software and autonomy — onboard model management, fault protection around autonomous decision-making, and FDIR philosophies that account for software that learns.
- RF and optical links — inter-satellite links and high-rate downlinks are the arteries of any distributed-compute constellation; phased-array and laser-comm engineers are already in short supply.
No single engineer covers all of that — which is exactly the point. Programs entering this space need teams, assembled fast, from a talent pool that was already tight before the orbital-AI wave arrived.
The hiring market this creates
We've watched this pattern before. When proliferated LEO constellations took off, RF and systems engineers with real satellite heritage became the most fought-over talent in aerospace, and programs that waited paid for it in schedule. The orbital-AI buildout is setting up the same dynamic — with an added twist: space primes and new-space startups are now competing for compute, power, and thermal talent against terrestrial data-center and semiconductor companies with deep pockets.
The constraint on the orbital AI boom won't be launch capacity or silicon. It will be the few thousand engineers on Earth who have actually flown hardware.
For hiring managers, the practical implications are blunt: senior satellite engineers with power/thermal, rad-tolerant electronics, FPGA, or autonomy depth will not sit on the market. Requisitions written for unicorns will age; programs that move decisively on strong 80% fits — and use experienced contractors to bridge the gap — will hold schedule.
Staffing for what's next
Fastwater Staffing is the number one space and satellite staffing firm for exactly this kind of moment. Our entire practice is senior contract engineers for space and defense programs — satellite systems, payload, RF and microwave, FPGA and ASIC, thermal and structures, mission analysis, and flight software — including the cleared and export-controlled talent (ITAR and Canada's Controlled Goods Program) that national-security space work demands.
As programs across the US, Canada, and Europe start adding compute-heavy payloads and autonomy to their roadmaps, we're already screening for the profiles they'll need. It's why clients treat us as the most trusted engineering staffing source for space and defense programs: when the market shifts, our bench shifts first.
The AI boom is going to orbit. The engineers who build it are already spoken for — unless you know where to find them.