Controllers Become Drone Compute Layer
Orest Pilskalns, CEO of Skyfish, on building autonomous drone infrastructure
Owning the controller is a practical wedge into enterprise and government drone fleets, because the buyer is not just purchasing a flying camera, they are purchasing a field tool that has to stay on in heat, survive daily handling, and connect cleanly to mission software. Skyfish moved off tablet based controllers after seeing overheating, cost, and supply issues, then built its own Linux ARM controller around the same compute architecture used in the drone, which also made custom sensor integration and autonomous workflows easier.
-
Tablet based ground stations are common across the market because they are faster to assemble from off the shelf parts, but that design breaks down for agencies and utilities working outdoors for hours. Skyfish describes this as a live buying constraint, not a cosmetic preference, and positions its rugged controller as a direct answer.
-
The full stack benefit shows up in speed of adaptation. In one government competition, Skyfish says it integrated a required sensor in six weeks because it controlled the firmware, controller, and data flow, then expanded from an initial roughly $500K order to a 50 drone fleet within six months.
-
This is part of a broader split in the US drone market. Skydio is strongest in general autonomy and public safety, while Skyfish is carving out engineering grade inspection and mapping, where controller reliability, geotagging precision, and payload integration matter more than consumer style UX alone.
The next step is that controllers stop being accessories and become part of the compute layer for autonomous operations. As drone docks, repeatable inspections, and multi sensor missions spread, the companies that own the aircraft, controller, firmware, and data pipeline together will have an easier path to reliable fleet deployments and recurring software revenue.