Google, NVIDIA and Emerald AI Launch Flexible Data Center Alliance
The new AI Energy Management Alliance aims to promote data centers that adjust their power consumption or use on-site energy resources in response to grid conditions.
AI Energy Management Alliance (AEMA) was established on September 16, 2026, at the initiative of Emerald AI, Google and NVIDIA. The coalition aims to promote an operating model for AI data centers in which large electricity consumers can reduce or shift their consumption depending on grid conditions, or use batteries and local energy generation.
The alliance comes at a time when the availability of electrical connections is becoming one of the constraints on AI infrastructure construction in the US. Its stated goal is to create conditions in which new large data centers do not operate only as fixed, continuous loads but can provide managed flexibility.
What the AI Energy Management Alliance wants to promote
AEMA says it wants to focus on measurable and verifiable requirements for demand flexibility. Its program also includes faster connections for large consumers through rules scaled according to the risk they pose to the power grid.
In practice, a data center could limit some computing tasks during periods of high grid demand, shift them to another time, or use its own batteries or local energy sources. The alliance thus brings together companies from the AI infrastructure and energy sectors around a common proposal for addressing growing data center demand.
According to Axios, the coalition has 20 companies and organizations. Its listed members include Anthropic, AES, National Grid, Constellation, NRG and RWE. AEMA builds on the earlier Advanced Energy Management Alliance, which focused on managing electricity consumption by large customers.
Data center connections are a key issue
Building large AI data centers requires substantial and stable electrical connections. AEMA wants to promote an approach in which a customer’s ability to temporarily adjust its demand is considered when assessing new connections. This could be particularly important in locations where grid capacity is limited or awaiting expansion.
The alliance also discusses the possibility of freeing up 100 GW of capacity, but this is a modeled estimate referring to an external analysis, not immediately available capacity. Likewise, claims about reducing electricity prices, preventing outages or significantly accelerating new connections currently represent goals and expected benefits, not independently demonstrated results from AEMA.
Rules and real-world operations will decide
The initiative’s significance will depend on specific regulatory rules and on data centers’ ability to reliably provide the promised flexibility. It has not been confirmed whether US regulators, transmission system operators and utilities will adopt the rules or incentives AEMA is seeking.
The next steps will include publishing the full list of members, the alliance’s governance structure and specific policy proposals. The responses of the federal regulator FERC, state regulators and regional grid operators will also be important. Ultimately, independent data from commercial deployments will determine the model’s practical viability—particularly the speed, scale and reliability of AI data centers’ demand reductions.
Sources
- NVIDIA Blog – Confirms the announcement of AEMA’s formation, its founding initiators and its stated principles for flexible connections.
- AI Energy Management Alliance – Confirms the organization’s existence and its stated purpose of promoting flexible AI data centers.
- Axios – Independently reports the number of organizations in the coalition, additional members and the context of the regulatory efforts.
- Heatmap News – Corroborates that this is a revived organization building on the Advanced Energy Management Alliance and reports on its leadership and focus.
Verified and updated: 09/16/2026 15:23
