
**TL;DR** - A stealth startup pitching satellite-based computation and encrypted mesh networking as a new layer of AI infrastructure.
1. Evaluate whether your inference workloads are latency-bound by terrestrial hops
2. Assess data residency constraints that make cloud region switching impractical
3. Request early access if running AI in contested or remote environments
**Example Prompt:**
Design a low-bandwidth model update pipeline for a satellite-connected edge deployment with intermittent connectivity.
| Pros | Cons |
|---|---|
| Truly novel infrastructure approach | Early stage, unproven at scale |
| Solves real data sovereignty problems | High latency to launch and maintain |
|---|---|
| Defense/enterprise TAM is large | Regulatory complexity is significant |