How “Power Couples” Can Help the United States Win the Global AI Race

RMI introduces the "Power Couples" model, which pairs large electricity consumers, such as AI data centers, with new-build wind, solar, and battery resources located near existing generators with approved grid interconnections. This approach aims to bypass long grid access wait times—which can reach five years for generation and eight years for transmission—while ensuring that the costs of new demand are borne by the users rather than general ratepayers. RMI's analysis indicates that this strategy could satisfy over 50 GW of new load at an average of 88% carbon-free energy, with significant opportunities concentrated in the PJM, ERCOT, and Southeast regions.

• The "Power Couples" strategy can rapidly satisfy over 50 GW of new data center or concentrated loads using an average of 88% carbon-free energy for less than $200 per megawatt-hour (MWh), with over 30 GW available at less than $100/MWh.
• US electricity demand forecasts have increased sharply; nationwide utilities expect demand to rise by 15.8% over the next five years, compared to a 4.7% expectation in 2023.
• Specific US utilities have seen dramatic upward revisions in load growth forecasts: Dominion Energy Virginia's expectation rose from 22% to 58% by 2035, Georgia Power moved from 0% to 48%, and Duke's forecast increased from 7% to 33%.
• Data centers are a primary driver of load growth, with McKinsey & Company projecting their power consumption will reach 606 TWh in 2030, a four-fold increase from 2023 levels.
• The highest potential for implementing Power Couples is located in the Southeast, ERCOT (Texas), and PJM (Mid-Atlantic), which are also the areas experiencing the heaviest forecasted load growth from manufacturing and data centers.
• The model utilizes "surplus interconnection" provisions from FERC Order 845 (2018) to allow new solar, wind, and storage to connect to the grid in a fraction of the typical wait time.
• To ensure equity and reliability, RMI proposes principles where the co-located load is physically protected to prevent it from burdening the grid if associated generation fails, and costs are assigned precisely to the co-locating parties.

APA

RMI (2025). How “Power Couples” Can Help the United States Win the Global AI Race. https://rmi.org/resources/how-power-couples-can-help-the-united-states-win-the-global-ai-race/

Chicago

RMI. How “Power Couples” Can Help the United States Win the Global AI Race. 2025. https://rmi.org/resources/how-power-couples-can-help-the-united-states-win-the-global-ai-race/.

Wikipedia

{{cite report |author=RMI |title=How “Power Couples” Can Help the United States Win the Global AI Race |date=20 February 2025 |url=https://rmi.org/resources/how-power-couples-can-help-the-united-states-win-the-global-ai-race/ |access-date=17 August 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{rmi2025how, author = {{RMI}}, title = {{How “Power Couples” Can Help the United States Win the Global AI Race}}, institution = {RMI}, year = {2025}, month = feb, url = {https://rmi.org/resources/how-power-couples-can-help-the-united-states-win-the-global-ai-race/}, urldate = {2026-08-17}, note = {Indexed by Climate Insights Directory} }