Google, Nvidia and startup Emerald AI on Wednesday relaunched an industry coalition around a simple but politically fraught promise: AI data centers will use less power when the grid needs relief.¹

The AI Energy Management Alliance, a revival of the largely inactive Advanced Energy Management Alliance founded in 2014, brings together 20 companies and organizations, including Anthropic, National Grid, AES, Constellation, NRG and RWE. Its narrower mission is to persuade utilities, grid operators and regulators to reward data centers that can flex their electricity demand.¹

For the companies, the pitch is a route through a growing bottleneck. Rather than operate as round-the-clock power consumers, data centers could shift or pause AI workloads, draw on batteries, or use nearby generation when supplies are tight. That could reduce the need for expensive grid upgrades and allow projects to connect faster. Google says it has committed 1 gigawatt of demand it can reduce through utility agreements nationwide.¹

Nvidia sustainability chief Josh Parker called the proposal a “win-win-win solution.”¹ Emerald AI chief executive Varun Sivaram set a tougher condition: any preferential access must rest on flexibility that is “verifiable and enforceable.”¹

The alliance is trying to answer a public backlash that its members cannot ignore. AP-NORC and the University of Chicago’s Energy Policy Institute found that 84% of Americans are concerned about data centers’ effect on local electricity prices; nearly equal shares of Democrats and Republicans favor requiring developers to pay for the grid upgrades they require.¹

That leaves the central test ahead. Federal regulators in June told regional grid operators to explore options for connecting large flexible users, opening a potential fast lane. But the industry must first convince regulators—and increasingly skeptical neighbors—that a “good grid citizen” is more than a slogan.¹