Grid software unicorn Emerald AI has teamed up with Google and Nvidia to form a coalition aimed at using the startup's technology to secure grid capacity for additional data centers. The newly established AI Energy Management Alliance (AEMA) seeks to integrate demand response—a strategy where data centers temporarily cut back on electricity usage—into the standard approach for data center development. By pausing noncritical operations and redistributing certain compute workloads, AEMA claims an extra 100 gigawatts of data centers could be connected to the grid.

EMBED_PLACEHOLDER_0

Joining the founding trio are Anthropic, along with several utilities such as AES, Constellation, National Grid, and NRG Energy.

Demand response is a longstanding concept that utilities have relied on for decades. It functions because the grid is built to handle peak loads. Throughout much of the year, demand stays well under the grid's maximum capacity. In the past, demand response primarily involved large electricity consumers like factories, which would agree to scale back their usage during high-demand periods—often by halting production or switching to backup generators. In return, utilities would compensate them, frequently quite generously. Nowadays, data centers can participate in these programs too, typically by activating backup generators.

Emerald AI has been advocating for a cleaner alternative to diesel generators. Its software synchronizes utility requests with data centers, enabling them to pause noncritical tasks or shift loads to other facilities where the grid has spare capacity. The startup isn't alone in pursuing this kind of technology. Google has been building its own solutions, and Enel X enables data centers to leverage their uninterruptible power supplies to smooth out peaks in power demand.

EMBED_PLACEHOLDER_1

Capping maximum grid usage at 90% for just a few hours at a time could free up 76 gigawatts of capacity for data centers, based on a study from Goldman Sachs released last year. Emerald AI, however, holds a key edge. Its software links utilities directly to data centers, which should let them respond rapidly to requests, functioning much like batteries. The startup also recently secured $150 million in a Series A round led by Energize Capital and DCVC, providing the funding necessary to expand the technology more broadly.

EMBED_PLACEHOLDER_2

Given the fluctuating nature of AI compute loads and data centers' capacity to quickly scale up and down, the formation of something like AEMA is long overdue. The coalition also aims to assist tech companies and utilities in identifying new locations for data centers—a task that has proven challenging for both sides. Despite its potential, more advanced demand response won't resolve all of the grid's current issues.