The U.S. Space Surveillance Network processes upwards of 25,000 observations daily, yet thousands of these detections remain classified as Uncorrelated Tracks. These unidentified objects—ranging from debris to potential adversarial maneuvers—often linger in the system due to the limitations of centralized, power-hungry computing. BQP aims to solve this by moving the intelligence directly to the edge.

At the core of the project is the company’s Physics-Constrained Quantum-Assisted Machine Learning (PC-QAML) software. By integrating physics-based modeling with quantum-inspired algorithms, the system shrinks AI model sizes by 99%, compressing 14 million parameters into just 2,000. This efficiency allows for deployment on resource-constrained hardware like the NVIDIA Jetson Nano, reducing power consumption by 90% and cutting inference latency tenfold.

Founder and CTO Rut Lineswala noted that the technology is designed for the harsh realities of space, where satellites must function with limited power and intermittent connectivity. Previous testing at the BMC3I TAP Lab demonstrated that this approach can autonomously detect separation events and orbital maneuvers, providing a faster alternative to traditional cloud-based processing. Following successful trials during the 2025 SDA Mini-Accelerator, this contract formalizes BQP’s role in supporting Space Operations Command and Space Systems Command objectives.