Emerald AI, Google and NVIDIA on September 16, 2026, announced the launch of the AI Energy Management Alliance (AEMA), a coalition aimed at advancing data centers that can dynamically manage their electricity use in response to conditions on the power grid.
The launch was announced in an NVIDIA blog post, which described AEMA as a first-of-its-kind coalition. NVIDIA stated that the objective is to build AI infrastructure that works with the grid rather than merely connecting to it, and said power flexibility can help unlock faster and larger connections for AI infrastructure while reducing environmental impacts per watt and supporting energy affordability.
The Case for Flexible Data Centers
NVIDIA stated that power has become a defining constraint on the expansion of U.S. AI infrastructure. Traditional interconnection processes, the company said, were designed around facilities with flat, static electricity demand, not computing infrastructure capable of responding when the power system is constrained.
A flexible data center can adjust the electricity it draws from the grid in several ways, according to the announcement: shifting computing workloads, discharging storage, using paired generation, or responding to system contingencies. Those capabilities allow a large electricity customer to serve as a controllable resource rather than an inflexible load. Used effectively, NVIDIA said, flexibility can make more efficient use of existing grid capacity, reduce demand during periods of system stress, avoid or defer infrastructure upgrades, and give utilities and grid operators greater confidence to connect AI facilities on shorter timelines.
Technology-Neutral, Performance-Based Principles
AEMA is technology-neutral and performance-based, focusing on the measurable service a facility can deliver, including response speed, duration, predictability and behavior during an emergency, rather than the specific hardware or software a site uses.
The alliance’s principles call for defining ride-through, curtailment and contingency-response obligations before a facility connects, which the announcement describes as clear rules for staying connected during brief grid disturbances, reducing power use when needed and responding to emergencies. The principles also call for standardizing technical requirements, performance metrics and operational data sharing, and for creating faster, risk-adjusted pathways for customers that make credible and verifiable flexibility commitments. A further principle addresses allocating interconnection costs in a way that reflects actual system impacts and benefits, such as avoided upgrades and improved ramping capability. According to the announcement, these measures can reduce uncertainty for developers while giving system operators the information and control needed to preserve reliability.
Alliance Scope and Policy Mission
AEMA convenes the full value chain across computing and power: AI platforms, infrastructure providers, data center operators, technology companies, power producers, utilities and regional grid operators. The founding members will be joined by launch partners from across the ecosystem, and the alliance will develop technical and operational approaches, collaborate with utilities on interconnection solutions and advocate for policies that recognize grid-responsive demand.
On its official website, AEMA lists Emerald AI, Google and NVIDIA as founding members and describes itself as an advocate for flexible data centers that dynamically adjust power draw to provide relief when the system needs it most. The alliance says it champions policies that accelerate speed-to-power and states that unlocking capacity on the existing, often underutilized grid protects energy reliability and affordability for American households.
AEMA states that half of the power system’s capacity goes unused throughout the year and that making data centers moderately flexible could unlock 100 GW from the existing power system, citing a Duke University Nicholas Institute publication. The alliance also cites a Brattle report finding that every 10% improvement in grid utilization can reduce utility rates by 3.4%, says demonstrations have reduced power consumption by a third in under a minute in emergency scenarios, and cites $733 million in potential avoided power-system costs for every gigawatt of new AI data centers with flexibility and capacity resources. AEMA states that connecting new AI data centers to the grid takes five to seven years or more in key U.S. markets.
The alliance says it maintains long-standing relationships with the Federal Energy Regulatory Commission, the Department of Energy, federal agencies and state regulators. It offers members policy coordination on interconnection, transmission and large-load policy, early insight into policy developments, and a unified message to state and federal regulators on the benefits of flexibility and speed-to-market, according to the site.
Prior NVIDIA and Emerald AI Collaboration
The alliance extends work already under way between two of its founders. In a March 23, 2026 press release issued at CERAWeek 2026, NVIDIA and Emerald AI said they were working with AES, Constellation, Invenergy, NextEra Energy, Nscale Energy & Power and Vistra to develop AI factories that connect to the grid faster and operate as flexible energy assets. Those factories would use the NVIDIA Vera Rubin DSX AI Factory reference design, which includes the DSX Flex software library for connecting AI factories to power-grid services, while Emerald AI’s Conductor platform orchestrates computational flexibility alongside onsite generation, batteries and other behind-the-meter resources. “AI factories are too valuable to be treated as either passive loads or permanent islands,” Varun Sivaram, founder and CEO of Emerald AI, said in the release.
According to the March release, the two companies trialed AI power-flexibility demonstrations at five commercial data centers around the world over the preceding year. DSX Flex was expected to be deployed at commercial scale later in 2026 at the NVIDIA AI Factory Research Center in Virginia, planned as one of the world’s first power-flexible AI factories running on Vera Rubin infrastructure.
NVIDIA’s launch post states that AEMA will broaden that earlier work by bringing the technology, energy and policy communities together around models that can be deployed across the U.S.
Credit: Source link
