Emerging Startup

Hardware Accelerators for Machine Learning Systems

Artist's rendering of network connections.

An emerging startup intends to commercialize cloud-based software to aid customers in performing AI and machine learning tasks. (Illustration credit: Shutterstock/Yurchanka Siarhei.)

Conventional computing platforms are inefficient at handling complex machine learning algorithms, which has limited the use of artificial intelligence (AI) in business applications. With support from Harvard's Physical Sciences & Engineering Accelerator, researchers in the labs of David Brooks and Gu-Yeon Wei are developing proprietary hardware to accelerate the pace at which machine learning algorithms can be processed. The product is cloud-based software to aid users in performing AI and machine learning tasks, at orders-of-magnitude faster speeds than existing commodity computation fabric. An emerging startup called Stochastic aims to offer an acceleration-as-a-service solution for businesses utilizing AI at enterprise scale.

Additional Information