AI data centers could reduce power draw on demand, study says
Source: Engadget
Overview
A recent UK trial demonstrated that AI data centers can dynamically adjust their energy demand without disrupting critical workloads, contrasting with the traditional always‑on power draw that can strain grids and increase prices.
Trial Details
- Period: Five days in December 2025
- Scope: More than 200 simulated “grid events” tested a London data center’s ability to modulate its energy use on the fly.
- Software: Emerald AI provided the control software.
- Partners: NVIDIA, National Grid, Nebius, and the nonprofit Electric Power Research Institute.
Results
- In every simulated grid event, the data center successfully adjusted its energy use to the requested level.
- Power draw was reduced by up to 40 % while critical workloads continued to run normally.
- The center responded to spikes in demand during soccer‑match halftimes, cutting power draw by 10 % for up to 10 hours.
- In one rapid response, the load was lowered by 30 % in just 30 seconds.
Future Plans
The study will serve as a blueprint for a 100 MW “power‑flexible AI factory” that NVIDIA plans to operate in Virginia.
“This trial proves that NVIDIA‑powered infrastructure can act as a grid‑aware asset, modulating demand in real‑time to support stability,” said Josh Paker, NVIDIA’s sustainability lead. “By making AI workloads responsive, we accelerate deployment while reducing the need for costly grid upgrades.”
Industry Implications
The organizations involved intend to share their data with the AI industry, regulators, and policymakers to influence future approaches. Reducing usage during peak demand could improve operators’ balance sheets and speed approvals for new data‑center grid connections.
“We would love to get to a point where we can get customers on the network in two years, and this is part of that,” said Steve Smith, president of National Grid Partners, to Bloomberg.
This article originally appeared on Engadget: https://www.engadget.com/ai/ai-data-centers-could-reduce-power-draw-on-demand-study-says-180628982.html?src=rss