During a tense soccer match between England and Germany, millions of Brits turned on electric kettles to make tea, causing a huge spike in electricity demand. But this time the grid was ready: an AI program instructed a London data center to slow down some power-hungry chips, balancing the load. This was a simulation conducted in December 2025 by engineers at Emerald AI, a Washington DC based firm, to test their software Conductor, designed to make data centers flexible.
Energy flexibility for data centers is an emerging solution to cope with the soaring electricity demand driven by artificial intelligence. Instead of building new power plants, which require years of permitting, data centers can temporarily reduce their consumption during peak periods, freeing up capacity. According to a Duke University study, this approach could unlock up to 76 gigawatts of additional capacity simply by reducing load for just 22 hours a year.
Sponsored Protocol
How Flexibility Works
Conductor, Emerald AI's software, will be deployed in a new facility in Virginia's Data Center Alley, connected to the live grid. When overall demand spikes, Conductor reduces power drawn by the chips while ensuring critical tasks still run. The project, in partnership with Nvidia and Digital Realty, is billed as one of the world's first "power-flexible AI factories." Josh Parker of Nvidia says flexibility is the bridge between the incredible demand for AI and the immediate limitations of our energy grid.
Not everyone welcomes data center growth. Over a dozen US states are considering bans, and localities like Minneapolis have imposed moratoriums. The GRID Act, a bipartisan Senate bill, would even sever new data centers from public grids. In this context, flexibility could improve public perception by reducing draw during stress times and decreasing the need for new fossil fuel plants.
Sponsored Protocol
A Princeton University report, funded by Google, found that a 500-megawatt flexible data center could reach full operation three to five years faster than an inflexible one. Flexibility not only speeds deployment but can also lower electricity costs and improve reliability. Johanna Mathieu, a grid expert at the University of Michigan, notes that demand flexibility helps reduce costs and improve grid reliability.
Sponsored Protocol
However, challenges remain. Traditional data centers are not designed for flexibility, and utilities are operationally conservative. Some skeptics warn that flexibility might distract from the need to build more grid infrastructure. Despite this, companies like Emerald AI, Nvidia, and Digital Realty are proving the concept in real-world settings. The upcoming deployment in Virginia marks a turning point.
To learn more about virtualization in data centers, read our article on HPE offering a free year of virtualization software. For background on the electrical grid, see the Wikipedia page on the electrical grid.
Source: https://www.technologyreview.com/2026/06/16/1138591/data-center-online-quickly-electric-grid-flex
