Ambient Scientific

Ambient Scientific is a fabless semiconductor company established in 2017 by CEO Gajendra 'GP' Singh, a seasoned professional with prior experience as VP of Engineering at Wave Computing. Operating out of Santa Clara, California, the company emerged from stealth mode in 2019 after securing $17 million in Series A funding. The firm focuses on the edge AI processor market, developing chips for AI inference and adaptive training on edge and embedded devices.

The company's business model is centered on designing and selling these specialized, ultra-low-power AI processors and associated development platforms to original equipment manufacturers (OEMs). Its clientele operates in diverse sectors such as automotive, healthcare, consumer wearables, and industrial applications. Revenue is generated from the sale of its AI System on Chips (SoCs), like the GPX10, and accompanying development tools, including a Software Development Kit (SDK) and a Neural Network Compiler.

Ambient Scientific's core offering is its proprietary DigAn® technology, an innovative architecture that combines analog and digital computing principles to achieve significant power efficiency for AI tasks. This technology enables the company's processors to perform AI computations, such as those for always-on voice detection or predictive maintenance, using minimal energy, comparable to that of a coin cell battery. The first product based on this architecture, the GPX10, is a fully programmable AI SoC manufactured by TSMC on a 40nm process. It integrates ten AI cores, an ARM Cortex-M4F control processor, and specialized hardware for sensor fusion, allowing it to process data from multiple analog and digital sensors simultaneously. The chip is designed to run without cloud dependency, enhancing data privacy and security. The company provides a full software stack that supports standard AI frameworks like Tensorflow and Keras, facilitating rapid application development for its customers.

The company's product line is expanding with offerings like the Sparsh-board and Sparsh AI Module, aimed at IoT, wearable, and smart device applications. The technology is positioned to enable a wide array of battery-powered, on-device AI applications, including driver monitoring systems, ECG wearables, always-on voice-based distress systems, and low-frequency computer vision.

Keywords: fabless semiconductor, edge AI, AI processor, ultra-low power, analog in-memory compute, DigAn technology, GPX10, on-device AI, programmable SoC, battery-powered AI, sensor fusion, AI inference, cloud-free AI, IoT applications, AI hardware, machine learning hardware, embedded systems, automotive AI, healthcare AI, industrial AI, consumer wearables AI, voice recognition chip, computer vision processor