Google calls Suncatcher a long-term research moonshot: can space one day host scalable machine learning infrastructure? The appeal is power. In the right orbit, a satellite sees near-constant sunlight with no night, weather, or atmosphere in the way. The research behind the mission is now peer-reviewed in the journal Joule. The prototype is not a data center. Reporting on the mission describes four TPUs on a single spacecraft, running short bursts of work to answer one question first: do commercial AI accelerators survive the space environment?
What the prototype is really testing
- Launch loads. Google shook the satellite on all three axes before flight. Individual components such as the chips see far higher shock than the vehicle-level load, so board-level mechanical design matters as much as the structure around it.
- Radiation. According to Google's published account, the team ran TPUs in a proton beam while running AI workloads and tracked how bit flips affected results. The ground data looked encouraging. Google says some things can only be tested in space, which is why this satellite exists.
- Heat. In vacuum there is no airflow. The only exit for heat is a radiator. Reporting on the prototype describes a stack of thermal interface material, aluminum and copper layers, heat pipes, and a radiator panel, with the chips running for minutes at a time before pausing to cool.
- Links. A future cluster would need high-bandwidth laser links between satellites flying in close formation. Google has described that as a separate, later mission.
A data center on Earth fights for grid power. A data center in orbit gets the power for free and then has to throw every watt of heat away through a radiator.
Why this is an engineering talent story
Every claim about orbital compute, from Google, SpaceX, and a growing list of startups, eventually runs into the same short list of hardware problems. None of them is a software problem, and none can be solved by hiring more machine learning engineers.
- Spacecraft thermal engineering. Radiator area scales steeply with power and temperature. Moving from tens of watts of avionics to kilowatts of accelerators per spacecraft turns thermal from a supporting discipline into the design driver. Engineers who have closed a thermal design on a high-dissipation payload, run thermal vacuum testing, and correlated the model to flight data are already scarce.
- Power systems. Large arrays, high-voltage distribution, battery sizing for eclipse, and the power electronics that feed dense compute boards. Senior power engineers who have taken a design through qualification sit at the center of every one of these programs.
- Radiation effects and reliability. Putting terrestrial-grade silicon in orbit means single-event effects testing, error mitigation in hardware and firmware, and honest analysis of what a bit flip does to a workload. Radiation-effects engineers have always been a small community. Commercial compute in orbit makes it busier.
- Board-level and FPGA design. High-speed digital boards that survive launch shock and thermal cycling, with the supervisory FPGAs and watchdogs that keep an accelerator recoverable when something upsets it.
- Optical inter-satellite links. The same pointing, acquisition, and tracking expertise that defense constellations are already competing for, now needed at shorter range and higher bandwidth for formation-flying clusters.
These are the same engineers that proliferated defense constellations, Earth-observation payloads, and space nuclear power programs are hiring. Orbital compute does not create a new talent pool. It adds another well-funded bidder to an existing one.
How smart programs will respond
Prototype programs move in campaigns: a thermal vacuum test, a radiation test window, a qualification run before a rideshare slot. Each needs senior specialists for months, not years. Teams that hold their schedules will keep a small permanent core and bring in contract engineers for the peaks, such as a thermal lead to correlate the model after first flight data or a radiation-effects engineer for a test campaign. Waiting two quarters for a permanent requisition to close means missing the next launch.
Where Fastwater comes in
Senior contract electrical and spacecraft engineering is the core of our practice. Fastwater Staffing is the number one staffing firm for the thermal, power, and radiation-hardened electronics engineers that orbital compute depends on, from power and harness leads to FPGA and board designers, thermal analysts, and the integration and test engineers who take hardware through thermal vacuum and out to the launch site. Our screeners ask how a thermal model was correlated or how a radiation test was structured, so the engineers who reach your team have built flight hardware.
We operate ITAR- and CGP-ready across the US, Canada, and Europe and place senior contract engineers in weeks rather than quarters. That is why program leaders call us the most trusted engineering staffing source for space programs racing from prototype to production.
Four chips are in orbit. Getting from four to thousands takes engineers who know how to move heat, power, and data in space, and we know where they are.