The grant, then the cost curve. On January 7, 2025, the University of Southern California was granted US12187462B2, "Using genetic algorithms for safe swarm trajectory optimization." A genetic algorithm is an optimization technique that evolves candidate solutions toward a goal; here the goal is planning safe, collision-free trajectories for a swarm of spacecraft. The classifications span B64G 1/242 (guidance), 1/1085 (orbital control), and G06N 3/126 (the genetic-algorithm AI class).
Now the segment economics, which is where this beat lives. Operating a satellite costs money in ground infrastructure and, critically, in people — flight controllers, conjunction analysts, mission planners. For a single satellite that is manageable. For a constellation of hundreds or thousands, hiring proportionally is ruinous. Autonomy is the only way the operating cost per satellite falls as the fleet grows.
That is the cost curve. Every piece of autonomy IP — collision avoidance, trajectory optimization, autonomous station-keeping — bends the operations-cost line away from linear and toward flat. A constellation operator with strong autonomy can add satellites without adding controllers; one without it watches its operating margin erode as it scales. Read the segment table of any large-fleet operator and the labor line is where this shows up.
The measured caveat: a university patent on a trajectory-optimization method is early-stage, and the path to it running live on a commercial constellation is not guaranteed. We track the direction of the autonomy cost curve, not a specific operator's adoption.
But the direction is unambiguous and it is the whole reason this IP keeps accumulating. The constellation business model only works if operating cost per satellite collapses as the fleet grows, and autonomy is the mechanism. A swarm-trajectory patent is, in financial terms, a claim on operating leverage — the most valuable thing a scaling space operator can have, and the hardest for a competitor to replicate quickly.