There are 43,000 satellites projected to be on-orbit by 2032. The hardware to receive their data hasn’t changed materially in twenty years. An estimated 98% of the data captured in orbit never reaches Earth.
That gap is a software problem, not a hardware one.
What we’re building
DeepBro is an AI software layer that spans the satellite-to-ground link — running onboard satellites and across existing ground station hardware (antennas, SDRs, RF chains, GSaaS platforms) — and uses edge AI to adapt to changing signal conditions in real time on both sides of the pipe. The platform starts with Automatic Modulation Recognition (validated on live satellite signals) and expands toward distributed antenna aggregation, blind demodulation, and RF environment intelligence.
Why now
Three things changed:
- Commodity SDR hardware is good enough to be a viable substrate.
- Edge AI accelerators (Jetson, equivalents) put sub-millisecond inference at the antenna.
- The economics of large monolithic dishes are increasingly hard to justify against a coordinated mesh of cheap nodes.
We think the next satellite-to-ground link is software-defined, deployed at the edge in orbit and on the ground, and cloud-managed. We’re going to write that software.
What’s next
- Live CatSat validation with FreeFall’s allsky antenna
- NSF Phase 1 SBIR submission (July 2026) on distributed antenna aggregation
- Pilot deployment with a partner ground station operator
If you operate ground stations, build satellite constellations, or invest in space infrastructure — we’d love to talk.