Rain Neuromorphics Secures Funding to Advance AI Chip Development

Article by: Sally Ward-Foxton

The startup’s neuromophic processing unit combines a training algorithm with a new analog AI chip architecture.

Rain Neuromorphics, the startup working on a full-fledged analog AI chip, raised $25 million in a Series A funding round. The company plans to invest the funds in product development while tripling its engineering and support staff.

The round was led by Prosperity 7 Ventures with existing investors Buckley Ventures, Gaingels, Loup Ventures, Metaplanet and Pioneer Fund, among others. Backers also include angel investors Sam Altman, co-founder and CEO of OpenAI, and Jeff Rothschild, founding engineer of Facebook. Altman previously led Rain Neuromorphic’s seed round in 2018.

Rain Neuromorphic Dendrite Diagram
Upper layers of the Rain’s neuromorphic processing unit. The vertical bitline columns serve as the “axons”, while the ReRAM devices are located at the interface between the columns and the randomly connected “dendrites”. (Source: Rain Neuromorphics)

The startups’ Neuromorphic Processing Unit (NPU) combines a new training algorithm – Equilibrium Propagation – with a new analog chip architecture. The combination can speed up processing and reduce power consumption, with an overall power reduction of a factor of 1,000 compared to current AI systems, the company claims. While analog computing is used today in some in-memory processor chips, the approach requires power-hungry ADCs and DACs between network layers. Analog computing in its current form is also incompatible with backpropagation, the widely used algorithm for training.

Rain’s NPU uses resistive RAM (ReRAM) as the memristive element, then combines it with 3D fabrication techniques and vertical bitlines borrowed from NAND flash technology. The approach allows Rain to create a chip modeled on the structure of brain cells. The vertical bitlines, coated with a memristive material, are analogous to axons, the CMOS layers below represent neurons, and the sparse, randomly patterned connections between axons and neurons are analogous to dendrites.

The point where a dendrite in any layer comes into contact with a column can be considered a synapse.

An NPU test chip recorded earlier this year is built on a 180nm CMOS process with 10,000 neurons. It has already demonstrated training and inference abilities.

Rain Neuromorphic’s long-term goal is to create a brain-like chip that can initially be used in cloud and edge AI applications.

This article was originally published on EE time.

Sally Ward Foxton covers AI technology and related issues for EETimes.com and all aspects of European industry for EE Times Europe magazine. Sally has spent over 15 years writing about the electronics industry from London, UK. She has written for Electronic Design, ECN, Electronic Specificer: Design, Components in Electronics and many others. She holds a Masters in Electrical and Electronic Engineering from the University of Cambridge.

Previous Members of the archery team learn a variety of skills at Koteewi Range • Current Posts
Next Nadroga players lack focus: Kumar – FBC News