Alphabet, Nvidia, and Emerald AI are asking utilities and regulators to treat AI data centers less like guaranteed round-the-clock electricity loads and more like customers that can reduce their draw when the grid is tight, tying that flexibility to faster grid connections and potentially lower infrastructure costs.
The companies on launched the AI Energy Management Alliance, or AEMA, alongside 18 launch partners spanning AI, utilities, power producers, storage, and grid software. Members include Anthropic, AES, Constellation Energy, NRG Energy, RWE, Fluence Energy, National Grid, and Voltus.
The alliance wants utilities and grid operators to create faster interconnection pathways for data centers that can verifiably reduce demand during constrained periods. Under its framework, facilities could shift computing workloads, discharge batteries, use colocated generation, or curtail electricity consumption when required.
Google has already moved partway toward that model. In March, the company said it had integrated 1 gigawatt of demand-response capacity into long-term agreements with multiple US utilities. Google said the system allows portions of machine-learning workloads to be shifted or reduced when electricity systems need relief.
Nvidia said AEMA will advocate for standards based on measurable grid performance rather than specific technologies.
“The result: faster AI deployment and a more efficient grid, without asking ratepayers to fund new capacity,” Nvidia sustainability head Josh Parker said.
The launch lands as the question of who pays for AI-related power infrastructure moves deeper into federal policy. On Wednesday, the US House passed the Ratepayer Protection Act 417-3 which would require state utility commissions to consider standards under which data centers drawing more than 100 megawatts would bear the incremental generation, transmission, and distribution costs needed to serve them.
Federal regulators are separately examining the flexibility-for-speed model. In June, the Federal Energy Regulatory Commission ordered six regional grid operators to justify or reform their large-load tariffs, including consideration of new transmission services for flexible customers and protections against cost shifting.
The Energy Department’s 2025 update estimated data centers could consume 11.8% of US electricity by 2030 in its reference case, with scenarios ranging from 9.5% to 15.3%.