Assessing the Additional Power Requirements of Artificial Intelligence in the United States by 2030
RAND Corporation's analysis of US energy system capacity to meet growing AI electricity demand — projections, bottlenecks, and recommendations.
Abstract
The RAND Corporation research study entitled "Assessing the Additional Power Requirements of Artificial Intelligence in the United States by 2030" analyses the ability of the US energy system to meet the growing electricity demand caused by artificial intelligence (AI) development. The study presents projections for potential increases in generating capacity on both the grid side (FTM — front-of-the-meter) and the consumer side (BTM — behind-the-meter).
The report proposes a new metric for more accurately comparing available supply with expected demand. This approach accounts for the significant gap between planned projects and actual capacity that becomes operational, addressing the "great shrinkage" phenomenon where ambitious plans face implementation challenges.
The authors conclude that, without significant changes to capacity expansion plans and the elimination of bottlenecks in the process of connecting new projects, a shortage of capacity is likely to occur by 2030. The report also provides recommendations for regulators and system operators to accelerate the integration of more reliable resources, such as energy storage and gas turbines.
Video Timeline
- 00:00 AI's Massive Appetite for Electricity
- 00:20 Can the US Power the AI Revolution?
- 00:54 AI's Insatiable Appetite: Projected Power Demand
- 01:36 The 1,000 GW Promise: New Power Projects Planned
- 02:21 The Great Shrinkage: From Plan to Reality
- 03:46 The Grid's Real Number: Net Available Capacity
- 05:01 Unlocking More Power: Potential Solutions
- 06:05 The Path Forward to Powering AI
RAND Corporation Report
Assessing the Additional Power Requirements of Artificial Intelligence in the United States by 2030
RAND Corporation Research Study
Read on RAND.org