\n\n
Key developments: Utilities and tech firms are expanding "demand response" programs that let data centers shift or curtail load during peak stress, a move seen as a way to avoid costly grid upgrades. The Electric Power Research Institute warns U.S. data‑center electricity use could jump from roughly 180 TWh this year to as high as 793 TWh by 2030, while surveys show facilities can trim peak demand by 10‑30 %.
Why it matters: Greater load flexibility could translate into $40‑$150 billion in avoided capital spending over the next decade, according to Duke University’s Nicholas Institute. Federal regulators are now drafting rules to fast‑track interconnection for flexible loads, and the AI Energy Management Alliance—led by Google, NVIDIA and Emerald AI—aims to standardize the approach. At the same time, Congress is debating legislation that would force new data centers over 20 MW to pay the full incremental transmission, generation and distribution costs they impose, reflecting growing concern that data‑center power use, now 4.7 % of U.S. electricity (up from 4.4 % a year earlier), could swell to nearly 12 % by 2030.
What to watch next: • Implementation of the new federal interconnection rules and how quickly utilities adopt flexible tariffs.
• Legislative progress on the bipartisan bill that would shift grid‑upgrade costs onto large data centers.
• Real‑world pilots, such as OpenAI’s pledge to cut up to 1 GW at its Georgia site during grid stress, and broader adoption of the AI Energy Management Alliance’s standards.
• Updates from Lawrence Berkeley National Laboratory on actual data‑center electricity use, which could confirm or revise the projected rise to 649 TWh by 2030. [[1, 2, 3]]