The Analog Processing Unit (APU) is a step-change in artificial intelligence.

This is made possible by the combination of Rain's proprietary algorithm and architecture.

The Analog Processing Unit (APU) is a step-change in artificial intelligence.

The world's first algorithm for end-to-end analog training and inference.

We collaborated with Turing Award winner Yoshua Bengio to develop an algorithm that allows our APUs to operate entirely in analog. Our algorithm is the world's first energy-based model realized as a physical circuit.





An architecture to scale up analog neural networks.

An architecture to scale up analog neural networks.

Sparse neural array

We have designed an architecture that scales in three dimensions to allow an unprecedented number of neurons and synapses to fit into a single chip. This architecture will enable neural networks to scale far beyond today's largest models, to ultimately reach brain-scale intelligence.

Our technology wins praise from the world's experts.

"Rain is...allowing extremely efficient, sparse connectivity as the basic building block of compute. Their chip allows for exploring models with much high dimensional states, closer to how the brain actually works."

Scott Gray

Technical Staff, OpenAI, formerly of Nervana Systems

"We're using the physics to directly implement the computations we want...that's why we can save so much in terms of computation time, energy, and size of the circuits."

Yoshua Bengio

2018 Turing winner, godfather of deep learning