Neural Magic, a US-based machine learning execution technology developer spun out of Massachusetts Institute of Technology (MIT), has received $15m in a seed round led by Comcast Ventures, the corporate venturing arm of mass media group Comcast, TechCrunch reported yesterday.
Enterprise software and services provider Amdocs also backed the round, as did venture capital firms New Enterprise Associates, Andreessen Horowitz and Pillar VC.
Founded in 2018, Neural Magic has begun early testing on a software development engine that exploits special algorithms to execute machine learning models on standard computer processing units, sparing the client expenditures on more powerful processing units such as GPUs or TPUs.
The technology harnesses the advantages of CPUs for machine learning – including more flexible computations and greater short-term memory availability – while maintaining raw processing outcomes comparable to advanced chipsets.
The spinout advances research undertaken by Nir Shavit, a professor in MIT’s Department of Electrical Engineering and Computer Science, together with his research associate Alex Mateev.
Neural Magic previously raised $5m of equity funding from undisclosed investors in March 2018, according to a regulatory filing.
Gil Beyda, managing director at Comcast Ventures, said: “Neural Magic is well down the path of using software to replace high-cost, specialised AI hardware.
“Software wins because it unlocks the true potential of deep learning to build novel applications and address some of the industry’s biggest challenges.”
– This article first appeared on our sister site, Global University Venturing.