AgenticScale launches distributed AI to support energy grid

AgenticScale has launched an approach to distributed AI infrastructure at the recent clean energy event, All Energy Australia.

Designed to grow Australia’s sovereign AI capacity while supporting the energy system that powers it, the approach places smaller, modular computing facilities at suitable energy and industrial sites, with capacity sized to the power available locally.

The company said it provides an alternative to concentrating AI computing in large data centres by bringing capacity to suitable power sources across a network of sites.

V2000, the latest addition to the company’s SAII Vault family, is designed around NVIDIA Vera Rubin computing technology to support this distributed approach.

The opportunity for energy companies is to make AI computing demand flexible. The design allows suitable workloads to be reduced or shifted when the local grid is under pressure, and scheduled when renewable power is available, within agreed customer service requirements.

AgenticGrid, AgenticScale’s electricity network digital twin, is intended to help identify suitable sites and understand local power constraints.

“AI infrastructure should be a good energy citizen,” said Karl Redenbach, Founder and Chief Executive Officer of AgenticScale. “We want to bring the right amount of computing capacity to the right locations and make its demand flexible enough to work with the local energy system.

“Australia can grow its AI capability in a way that supports energy infrastructure. The opportunity is to plan computing and power together, so that new AI capacity can be a positive part of the energy system.”

AgenticScale has sites going live in Australia as it builds out its distributed AI network.

Image credit: iStock.com/AlexSecret. Image used for illustrative purposes only.