The coming decade will see an explosive increase in data generated by edge devices such as connected vehicles, smart city cameras, and drones. Yet wireless network bandwidth will remain a critically limited resource. This creates an urgent need for AI edge devices to determine which data is relevant, build efficient representations, and transmit for cloud services, thereby minimizing network transmission and data storage costs.
Our vision is to develop a purpose-built AI in-memory-computation processor and accompanying software platform to address these problems at 10x greater efficiency than the state-of-the-art, transcending wireless bandwidth limitations and enabling low latency services at mobile power budget. The core technologies were developed under DARPA contracts and are now in the process of commercialization.