UK Trial Shows Data Centers Can Reduce Electricity Demand with AI Technology
A UK trial led by National Grid, in partnership with Emerald AI, EPRI, Nebius, and NVIDIA, demonstrated that data centers can adjust electricity consumption using AI. The trial at a London data center utilized Emerald AI's software to manage 96 NVIDIA Blackwell Ultra GPUs, achieving a 40% reduction in demand during simulated grid events. The technology effectively responded to demand spikes and maintained lower power use, showcasing potential benefits for grid stability and renewable energy integration.

A UK trial demonstrated that data centers can modify electricity consumption in response to grid signals using AI technology. Led by National Grid with partners Emerald AI, EPRI, Nebius, and NVIDIA, the trial took place at a London data center using Emerald AI's 'Emerald Conductor' software to manage 96 NVIDIA Blackwell Ultra GPUs.
The system reduced electricity demand by up to 40% during over 200 simulated grid events. It also lowered load by approximately 30% within 30 seconds during stress tests. This approach may help reduce grid constraints and support renewable energy connections, allowing for more efficient management of electricity demand as the digital economy grows.




Comments