Nvidia (NVDA) and Google have formed a coalition aimed at slashing AI data center energy consumption.

Nvidia, Google, and energy technology firm Emerald AI have announced a partnership centered on electrical power management, unveiling the AI Energy Management Alliance (AEMA) on Wednesday. The coalition’s primary objective is to develop AI infrastructure capable of dynamically adjusting electricity consumption in response to grid requirements. By transforming traditionally static data centers into adaptive facilities, the alliance aims to shift computing tasks to periods of lower demand, utilize battery reserves, or temporarily decrease power draw during peak grid strain. To standardize these operations, AEMA will establish unified benchmarks for response times, dependability, and operational metrics.

Under the framework outlined by Nvidia, adaptive data centers will deploy multiple strategies to navigate grid constraints while maintaining a technology-agnostic approach. Rather than mandating specific hardware or software platforms, the alliance will focus exclusively on measurable outcomes. Prior to Wednesday’s official launch, Nvidia and Emerald AI conducted preliminary research demonstrating that lower-priority computational processes could be safely suspended or rescheduled without disrupting critical AI workloads. This adaptability significantly streamlines the approval process for utility companies seeking to connect new AI infrastructure. Grid management teams will gain greater assurance when authorizing substantial power allocations, knowing that participating facilities possess the operational capability to scale back when necessary.

The alliance’s formation coincides with a critical inflection point for global electricity systems. According to the International Energy Agency (IEA), worldwide data center electricity consumption climbed 17% in 2025, totaling approximately 485 terawatt-hours. Facilities dedicated specifically to AI processing experienced an even steeper 50% surge in energy demand. IEA projections indicate that total data center power usage could approximately double before 2030 ends, climbing to roughly 950 terawatt-hours. At that level, data centers would account for approximately 3% of global electricity consumption. Within the United States, the pressure on local grids is particularly pronounced; data centers contributed to roughly half of the nation’s total electricity consumption growth throughout 2025.

To address these systemic challenges, AEMA intends to unite stakeholders across the AI and energy sectors, encompassing AI technology providers, data center management firms, electricity generators, utility organizations, and regional transmission operators. The coalition will also collaborate directly with utility providers on infrastructure connection strategies and advocate for regulatory frameworks that formally acknowledge and incentivize demand flexibility. Following Wednesday’s announcement, shares of both Nvidia and Alphabet experienced modest gains at market opening. Building on the successful real-time grid response trials previously conducted by Nvidia and Emerald AI, AEMA plans to scale this adaptive infrastructure model throughout the United States.