Published evidence log
What energy sources will supply the AI boom
Current position
Working answer
The AI boom will likely be supplied by a diversified electricity system rather than a single energy source. Existing grid resources—natural gas, nuclear, hydro, solar, wind, coal, and limited oil/diesel—can all contribute, while new generation, storage, and transmission are built around data-center demand. The source highlights natural gas as a flexible option and solar and wind as low-marginal-cost sources, but says the winning mix will depend heavily on location, contract length, and grid infrastructure.
The source frames the AI boom primarily as a power-demand and infrastructure problem rather than naming one winning fuel. AI data centers already represent about 5% of U.S. power consumption, with demand described as doubling roughly every two years. Meeting that growth will require electricity from the existing grid mix—solar, wind, hydro, nuclear, natural gas, coal, and smaller oil/diesel generators—as well as new generation, storage, and transmission. Natural gas is described in particular as a dispatchable generation technology, while solar and wind are identified as low-marginal-cost renewable sources. The source's central conclusion is that AI demand can justify new generation, but the commercially viable mix will depend on location, contract duration, and surrounding grid infrastructure. It therefore supports a diversified answer: renewables and nuclear can provide generation, gas can provide flexible supply, and the grid and storage determine whether those resources can reliably serve data centers.
Chronology
Evidence timeline
August 5, 2026
Power 2026 - Electricity Pricing in the Age of AI[1]
The primer argues that power—not GPUs or memory—is becoming the key constraint on expanding AI capacity. It says data centers already account for approximately 5% of U.S. power consumption and that data-center demand is doubling every two years. It describes the U.S. generation mix as including solar and wind, along with hydro, nuclear, natural gas, coal, and oil/diesel units. The source explains that natural-gas plants generate electricity by burning gas to drive a turbine, while other generation technologies use different mechanical or photovoltaic processes. It concludes that AI demand can support new generation, but the economics depend on contract duration, location, and surrounding infrastructure.
How this bears on the question
This primer is directly relevant because it explains both the scale of AI-related electricity demand and the generation technologies that could meet it. It does not forecast a single dominant fuel; instead, it presents AI demand as a driver of new generation and stresses that project economics and reliability depend on local transmission, contracts, and infrastructure.
Bibliography
Sources
- 1.