Tech giants join coalition to use AI for grid capacity for data centers
A new alliance called the AI Energy Management Alliance includes Google, Nvidia, Anthropic, and utilities, using Emerald AI's software to coordinate demand response. The goal is to free up 100 gigawatts of grid capacity by having data centers temporarily reduce power use. This could allow more data centers to connect without overloading the grid.
Demand response has long been a utility staple, with factories historically paid to pause production or switch to backup generators during peak grid stress. Emerald AI's approach differs by using software to link utilities directly with data center operators, enabling rapid load-shifting between facilities where grid capacity exists. The startup's recent $150 million Series A from Energize Capital and DCVC positions it to scale this coordination model.
The coalition's 100-gigawatt target is ambitious but not without precedent—Goldman Sachs estimated that capping grid usage at 90% for brief periods could unlock 76 gigawatts. However, Emerald AI's chief scientist acknowledges the technology blunts, rather than eliminates, the need for new generation sources. Google and Enel X are pursuing similar demand-response tools, indicating a competitive landscape.
This alliance could reshape how data center growth intersects with grid reliability, potentially accelerating AI infrastructure expansion without requiring massive new power plants. Communities facing data center moratoriums due to grid constraints may see development resume, while utilities gain a flexible tool to manage peak demand. Ratepayers could benefit from deferred infrastructure costs, though the effectiveness depends on data centers' willingness to interrupt workloads—a tradeoff that may affect service reliability for cloud customers. The approach could also reduce reliance on diesel backup generators, with localized environmental benefits.