AI Chip Startup Syntiant Scales Production
Syntiant Corp., the “neural decision processor” startup, announced completion of another funding round this week along with the shipment of more than 1 million low-power edge AI chips.
The three-year-old startup based in Irvine, Calif., said Tuesday (Aug. 4) its Series C funding round generated $35 million, raising its venture capital total to just over $65 million, according to the web site Crunchbase.com.
The round was led by Microsoft’s (NASDAQ: MSFT) venture arm M12 and Applied Ventures, the investment fund of Applied Materials (NASDAQ: AMAT). New investors included Atlantic Bridge Capital, Alpha Edison and Miramar Digital Ventures.
Intel Capital was an early backer of Syntiant, part of a package of investments the chip maker announced in 2018 targeting AI processors that promise to accelerate the transition of machine learning from the cloud to edge devices.
Syntiant’s speech-recognition processors are used in battery-powered edge devices ranging from smartphones and smart speakers to laptops and earbuds. They are also used in sensor platforms. Early use cases include “wake” and command words, speaker identification and event detection.
The startup claims its processors offer a 100-fold improvement in power efficiency and a throughput improvement over microcontrollers and digital signal processors. The AI processor combines silicon with data collection and deep learning training of its parallel neural network.
Syntiant’s neural processors incorporate deep learning algorithms into its chip design used in low-power neural computers integrated with battery-powered devices. The startup’s NDP 100 processor measures 1.8 mm by 1.4 mm, and consumes 140 uW when operating.
While the speech and voice recognition market has been forecast to approach $27 billion by 2025, chip industry suppliers such as Applied Materials are seeking to expand low-power processor technologies as tools for sensor applications. Applied Materials said backing Syntiant was part of its investment strategy for transitioning semiconductor materials to systems.
Among the potential use cases of AI processors is an emerging discipline called “materials informatics” that uses data analytics techniques to advance materials science R&D. One concept known as “inverse design” involves model training using material properties and processing steps as data points.
According to report by IDTechEx, “The technology is applicable to anyone that designs materials or designs with materials, and [the] aim is to have this inverse design fully integrated with initial product design.”
Syntiant said Michael Stewart, Applied Ventures’ investment director, will join its board of directors.