femtoAI

The company says its first-generation Sparse Processing Unit (SPU) aims to combine ASIC efficiency with the flexibility of a general-purpose accelerator. The page says it supports programming, simulation, and deployment.

The page presents production-ready AI models optimized for efficiency, particularly when used with the SPU. Its pre-trained models use diversified datasets, and its pre-optimized models are pruned.

The page describes AI noise reduction for hearables, including speech intelligibility, adjustable noise passthrough, reduced noise exposure when paired with ANC, and low-latency, low-power models.

Femtosense offers an embedded AI platform with pre-trained models, an SDK for custom applications, and a Sparse Processing Unit accelerator.

Emerging AI processing technology developed by Femtosense will bring hearing aids to the level of performance, compact form factor, and ultimately, cost required for mass adoption of hearing aids.

The page describes expectations for intelligent, conversational devices and the need for smart glasses to provide always-on speech recognition for commands, translation, and context-aware assistance.

The page describes voice as a human interface in products from appliances and wearables to industrial equipment, with deep neural networks used to recognize commands and wake words and separate speech from noise.

The page describes how portable devices can run AI workloads locally within low-power budgets, with fast response times and constrained form factors.

The page invites visitors to meet femtoAI at CES 2026 and includes a meeting request form.

ABOV and femtoAI to lead edge inference into the future with new 5 year deal

femtoAI announced 5x growth in the first half of 2026, driven by engagements with Samsung and Marshall. The release says its dual sparsity delivers 100x power and energy gains and uses 10x less memory.

The careers page describes femtoAI's embedded AI solutions on its Sparse Processing Unit platform and invites applicants to join the team.

An embedded systems engineer at femtoAI develops firmware for its Sparse Processing Unit and system-on-chip platforms, including RISC-V management processors and firmware infrastructure.

femtoAI develops deep learning solutions on its Sparse Processing Unit (SPU) platform for devices such as earbuds and cameras.

femtoAI (formerly Femtosense), founded in 2018 by Stanford University's Brains in Silicon Lab researchers, is headquartered in Silicon Valley. Its neuromorphic technology uses sparsity and includes an AI accelerator for end-to-end embedded AI.

femtoAI, formerly Femtosense, was founded in 2018 by researchers from Stanford University’s Brains in Silicon Lab and is headquartered in Silicon Valley. Its technology draws on neuromorphic computing principles such as sparsity.

Embedded AI platform built upon a unique AI accelerator capable of delivering higher performance at lower latency, using less energy in smaller silicon.

Embedded AI platform built upon a unique AI accelerator capable of delivering higher performance at lower latency, using less energy in smaller silicon.

The privacy policy explains how FEMTOSENSE collects, uses, stores, and discloses personal data, including data collected through website visits, newsletter sign-ups, purchases, and customer support.

The FEMTOSENSE privacy policy explains how the company collects, uses, stores, and discloses personal data, including data collected through website visits, newsletter sign-ups, purchases, and customer support requests.

The End-User License Agreement, last updated May 13, 2024, covers Femtosense software development kits and the femtocrux, femtodriver, femtoflow, and fmot software.

femtoAI's policy outlines its commitment to ethical business practices and efforts to address human rights abuses, child labor, and armed-conflict financing in the supply chain for critical and conflict minerals.

femtoAI Standard Terms and Conditions Feb-2-2026Download

GitHubRepository owner

femtoAI provides open access to its developer community to create AI-powered products using its full-stack AI silicon and compression platform. The company aims to accelerate innovation by enabling developers to build applications leveraging its hardware and software solutions. femtoAI’s platform supports end-to-end AI workflows, from edge devices to cloud systems.

Oct 2026PR NewswireNews article

femtoAI reports 5X growth in the first half of 2026, driven by customer engagements around its full-stack sparse edge AI inference platform and Sparse Processing Unit. The announcement presents the platform and unit as the focus of those engagements, but gives no further details about the customers or how growth is measured.

Jul 2026PR NewswireNews article

The piece analyzes femtoAI’s sparse-silicon approach to running AI inference efficiently at the extreme edge.

Jun 2026The ThesisNews article