Study shows promise for ‘flexible demand’ data centres
Artificial intelligence (AI) data centres could be turned into power-flexible assets, harnessing AI to adjust power consumption in real time, a new study has found. The live trial, a UK-first, involved the National Grid, Emerald AI, EPRI, Nebius and NVIDIA[i].
Over the course of five days last December Emerald AI’s software ‘Emerald Conductor’ was used on a cluster of high-performance NVIDIA GPUs at Nebius’s new data centre in London. During the test more than 200 real-time simulated “grid events” were sent to the site to assess how Emerald’s software would respond. Each time the AI platform successfully adjusted power use to the requested level, cutting demand by up to 40% while critical workloads continued to run as normal.
The trial also demonstrated that the system could respond to sudden demand spikes, follow load-reduction requests for extended periods, and rapidly reduce electricity use during simulated system stress events. In one scenario, the system cut around 30 % of its load within roughly 30 seconds.
It validates that artificial intelligence data centres can dynamically adjust power consumption in response to real-time signals, without disrupting critical workloads. Given that most large data centres today are fixed and “always on”, it means that by applying AI, data centres can move from being a source of electricity constraint to a controllable grid asset. By flexing demand in real time, they can help manage peaks, make better use of existing infrastructure, and support the connection of different sources of energy to the grid.
Steve Smith, President, National Grid Partners commented:
“As the UK’s digital economy accelerates, there’s concern that data centres could add pressure to an already constrained system. This trial proves the opposite can be true. High‑performance data centres don’t have to place additional strain on the grid. With our partners, we’ve shown they can be connected and managed without major new network capacity, flexing their power up or down in real time to support the whole system. This approach will enable us to connect significant new demand more quickly and, help to lower network charges for customers over time.”[ii]
The researchers explain that the UK is currently preparing for more than 6 GW of data centre deployments on the grid by 2030. Using the technology applied in the study could enable AI data centres across the UK to add more than 2 GW of capacity back to the grid when required. Dr Varun Sivaram, Founder and CEO of Emerald AI, adds:
“This trial demonstrates that AI infrastructure can be a dynamic force for the grid. With dozens of realistic AI workloads running simultaneously, we delivered fast emergency curtailment and sustained, precise peak reduction. The same approach we validated here can be applied to much larger AI factories, as the industry scales.”[iii]
References
[ii] Ibid
[iii] Ibid
