femtoAI

femtoAI develops ultra-efficient AI inference technology for the extreme edge, enabling AI to run directly on everyday devices without cloud connectivity. Founded in 2018 and headquartered in Silicon Valley, the company was spun out of Stanford University's Brains in Silicon Lab and applies a neuromorphic-inspired engineering approach. It was previously known as Femtosense and rebranded to femtoAI in 2024. Total funding raised stands at $20.6M.

At the centre of the company's full-stack offering is the Sparse Processing Unit (SPU-001), custom sparse processing hardware that achieves 100x energy efficiency and 10x model capacity compared with traditional AI processors, making AI viable on battery-powered and cost-sensitive devices. The hardware is paired with a Software Development Kit and ready-to-deploy AI models, which together deliver up to 100x power consumption reduction and 10x memory footprint reduction with minimal performance impact, enabling real-time AI on compact silicon.

Technical work spans edge AI inference, neuromorphic computing, sparse processing, custom AI hardware and silicon, AI model optimisation, and embedded software and SDK development. Target industries include consumer electronics, hearing aids and hearing health, smart home devices, wearables, robotics, and autonomous vehicles.

The technology is already shipping in real-world products, among them the NewSound AI over-the-counter hearing aid. femtoAI's recent recognition includes a CES 2026 Best-in-Show award, the Startup Terrace Award at InnoVEX 2025, and membership of the NVIDIA Inception Program.