Google has announced implementation agreements with Indiana Michigan Power (I&M) and the Tennessee Valley Authority (TVA) to introduce Flexible Demand or Demand Response capabilities at its US data centers. This allows for the reduction or shifting of AI-related power consumption during specific periods at the request of grid operators, alleviating pressure on the power system amidst growing demand. The significant increase in AI power load drives these measures, covering data centers in Fort Wayne, Indiana, and the TVA service area, serving as a key strategy for the tech giant to ease system congestion.

Google conducted pilot testing in Omaha, Nebraska, reducing demand for machine learning processing loads during three grid stress events in 2024. The technical approach involved rescheduling non-urgent computational tasks (such as video processing) during critical periods for the power system. This helps data centers connect to the grid faster and manage peak demand resources efficiently. Agreement constraints include: limiting non-essential AI load activities (like model training or data analysis) only during demand surges or extreme weather events, rescheduling processing times or locations, without affecting cloud services for sensitive sectors like search, maps, or healthcare. I&M executives emphasized that large consumer flexibility is crucial for efficiently planning generation and transmission infrastructure, while reducing the urgent need for immediate investments in the grid and power plants.

Furthermore, Google plans to expand this capability to other regions, drawing on existing experience from Belgium and Taiwan. Flexible Demand is in its early stages for data centers, with deployment dependent on operator agreements, system design, and service reliability requirements. AI model training can consume tens to hundreds of megawatts (MW) over extended periods, making optimizing energy use and coordinating with the power system critical for managing new facility expansion. These agreements complement Google's investments in other technologies like hydropower and nuclear power, forming part of its solution portfolio for data center power growth amid rapid AI penetration.
url:https://www.pv-magazine-latam.com/2025/08/06/google-implementa-acuerdos-para-flexibilizar-el-consumo-energetico-de-centros-de-datos-con-ia-en-estados-unidos/