Pentagon seeks ultra-fast AI to fuse missile and space sensor data

The U.S. Pentagon, through its Defense Innovation Unit, has launched a solicitation for an artificial-intelligence platform capable of synthesising disparate sensor feeds—live video, satellite pictures, radar returns and classified reports—into a continuously refreshed picture of space and missile threats. The system must handle regular data flows of about 20 to 30 megabytes per minute and occasional bursts up to 5 gigabytes, delivering actionable, confidence-scored alerts with a latency no greater than five seconds, with two seconds as the preferred target.

Existing solutions struggle with distinguishing tightly spaced objects and maintaining current threat models, prompting the Pentagon to seek AI that can process and correlate information in real time. The proposed engine will support analyst-in-the-loop, human-on-the-loop, or automated workflows, providing operators with clear visualisations while preserving human oversight of critical decisions. Faster processing aims to give commanders extra time to assess incoming missiles or drones and launch interceptors. The initiative reflects growing data volumes from modern missile and drone attacks and the need to shorten the decision-making cycle in missile defence.