1. Home
  2. Companies
  3. femtoAI

About

femtoAI, founded in 2018, is a technology company originating from Stanford University's Brains in Silicon Lab. It develops ultra-efficient AI inference technology designed for the extreme edge, enabling artificial intelligence to run directly on battery-powered and cost-sensitive devices without requiring cloud connectivity. The company is based in Silicon Valley.

The company's core technical domains include sparse processing, neuromorphic computing, and energy-efficient hardware design. Its product offerings center on the Sparse Processing Unit (SPU-001), a hardware chip that achieves 100x energy efficiency and 10x model capacity compared to traditional AI processors. A full-stack solution combines this custom hardware with deployable AI models and a software development kit (SDK) for real-time applications on compact silicon.

femtoAI's technology is shipping in real-world products, including the NewSound AI over-the-counter hearing aid. The company has raised $20.6 million in funding and operates within industry verticals such as consumer electronics, wearables, robotics, and autonomous vehicles. It has received recognition through awards including the CES 2026 Best-in-Show and the Startup Terrace Award at InnoVEX 2025, and is part of the NVIDIA Inception Program.

Similar companies

LA

Lyte AI

Lyte AI develops end-to-end vision systems for robotics and autonomous vehicles, enabling machines to perceive and navigate the physical world using 4D edge AI processing.

1 job
Alif Semiconductor logoAS

Alif Semiconductor

Alif Semiconductor develops AI-enabled microcontrollers and embedded processors for edge computing, IoT, automotive, and industrial applications.

EI

Etched.ai, Inc.

Etched.ai develops transformer-specific ASIC chips designed to accelerate AI model inference with higher throughput and lower latency than general-purpose GPUs.

Cerebras logoCE

Cerebras

Cerebras builds wafer-scale AI supercomputers that consolidate the computational power of dozens of GPUs onto single chips for accelerated machine learning training and inference.

Extropic (Extropic Corp / Extropic AI) logoE(

Extropic (Extropic Corp / Extropic AI)

Extropic develops energy-efficient computer hardware using thermodynamic computing to process probabilistic AI workloads.

Normal Computing logoNC

Normal Computing

Normal Computing builds AI-accelerated software and physics-based hardware to improve efficiency in AI infrastructure and semiconductor design.