Google, Nvidia and Emerald AI launch alliance to fix the power problem making AI data centers unpopular

- Google, Nvidia and Emerald AI launched a coalition to promote flexible power use by AI data centers.
- The alliance wants utilities to reward data centers that cut power demand during grid stress.
- The coalition says flexible AI facilities could unlock faster grid connections and reduce costly infrastructure upgrades.
Google (NASDAQ: GOOG), Nvidia (NASDAQ: NVDA), and the startup Emerald AI have launched a new coalition that wants regulators to hand faster grid connections to data centers that are willing to dial back their power use when supplies get tight.
The coalition, called the AI Energy Management Alliance, was announced on Wednesday, September 16.
The alliance arrives at a time when there are increased calls for data center projects to get scrapped or paused, with rising electricity bills and environmental factors being some of the leading grievances.
A relaunch built around one narrow idea
The alliance is being seen as a revival. An older body, the Advanced Energy Management Alliance, was set up in 2014 to lobby for demand-response firms and had gone dormant, Emerald AI CEO Varun Sivaram told reporters.
The mandate of the latest coalition is not as broad as its predecessor. This time around, the focus is to convince utilities and regulators that flexible AI data centers are assets worth rewarding.
The coalition reportedly opened with 20 companies and organizations from the AI and energy sectors. Among the members are the AI lab Anthropic, utilities National Grid and AES, and power producers Constellation, NRG, and RWE.
Before the official announcement of the launch, Nvidia’s head of sustainability, Josh Parker, said, “It’s really a win-win-win solution.”
According to Sivaram, their aim is to turn data centers into “good grid citizens” that a town sees as an asset instead of a burden.
What is the bargain?
Data centers are viewed by utilities as heavy power consumers who need a constant supply. However, the coalition wants to change that assumption and show that facilities can reduce demand during periods of high stress when supply is tight. In return, facilities that are able to do this, that is, reduce their demand, will be able to connect sooner back and, in some cases, dodge expensive grid upgrades.
That flexibility can come from pausing or shifting AI computing jobs, drawing on batteries, or leaning on nearby generation.
Google’s head of advanced energy market innovation, Tyler Norris, said the company has pledged 1 gigawatt of demand it can cut through utility agreements.
Sivaram says that modest flexibility during the grid’s worst hours could free up 100 gigawatts on the existing system, which he said runs only about 50% utilized on average.
The group conceded that the faster connection deal only works if the promised flexibility is real. According to Sivaram, the facilities should only qualify if they have proven that they are capable of cutting demand, and it must be verifiable.
Emerald AI and Nvidia say they have run six flexibility demonstrations globally. Later this year in Virginia, Nvidia, Digital Realty, and Emerald AI plan to switch on a power-flexible AI facility of nearly 100 megawatts.
Federal regulators gave the approach an opening in June, when the Federal Energy Regulatory Commission (FERC) told regional grid operators to weigh new ways of connecting large, flexible customers.
Why launch a coalition now?
Public opinion on data centers is not as favorable as those investing in them would want. New AP-NORC and University of Chicago Energy Policy Institute polling released Wednesday found that 84% of Americans are worried about data centers’ effect on local electricity prices, and 53% are now extremely or very concerned about AI’s environmental impacts, up from 41% a year earlier.
Support for making developers pay for grid upgrades crosses party lines, at 79% of Democrats and 76% of Republicans.
Ben Green, a University of Michigan researcher, said that communities have organized effectively enough to block or pause many projects, calling the local economic benefits a “significant false promise.”
In early August, Greg Abbott, the Governor of Texas, paused for new data centers until there was a comprehensive audit of those who wanted to connect to the state’s electric grid, and it is not unrelated to the grievances of the citizens who have shared concerns about usage and costs.
Lawmakers are moving too. The US House may vote this week on the bipartisan Ratepayer Protection Act, which would let states force data centers drawing more than 100 megawatts to fund their own power infrastructure.
Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.
FAQs
What is the AI Energy Management Alliance?
It is a coalition founded by Google, Nvidia and Emerald AI, launching with about 20 members including Anthropic, National Grid, AES and NRG, that campaigns for regulators to reward AI data centers that can flexibly reduce their grid demand.
How would flexible data centers help the grid?
By shifting or pausing computing jobs, using batteries or tapping nearby generation, facilities can cut demand during peak hours; Emerald AI's Varun Sivaram argued in Fortune that even modest flexibility could unlock 100 gigawatts on the existing grid, which runs about 50% utilized on average.
Why is public opposition to data centers rising?
AP-NORC/EPIC polling released Wednesday found 84% of Americans concerned about data centers' effect on local electricity prices and 53% concerned about AI's environmental impacts, up from 41% a year earlier, with support for making developers pay for grid upgrades spanning both parties.

Hannah Collymore
Hannah is a writer and editor with nearly a decade of blog writing and event reporting experience in the crypto space. At Cryptopolitan, Hannah contributes to the news page, reporting and analyzing the latest developments in DeFi, RWA, crypto regulation, AI and frontier tech industries. She graduated from Arcadia university with a degree in Business Administration.
















