Claim report cards
Are the common AI cost claims true? Five rulings
5 rulings, one per claim: true that AI drives the return of U.S. electricity-demand growth and true that it is making memory and storage more expensive; false that AI caused the transformer shortage or that every prompt uses a bottle of water; unproven that AI raised any one household's electricity bill by a specific amount.
4 of 5 claims are settled outright by the cited evidence. The remaining one is unproven: no published measurement supports them in either direction.
Which claims are true, false, or unproven?
“AI is driving the return of U.S. electricity-demand growth.”
AI data centers are the largest single driver of new U.S. electricity demand.Data centers are the largest new source of U.S. electricity demand and are projected to use 11.8% of national electricity by 2030 in LBNL’s reference case (range 9.5% to 15.3%, 2026), up from about 4.4% in 2023. EIA names data-center load as the cause of current load growth. National figures do not separate AI from ordinary cloud workloads facility by facility.
Open report card“AI raised my household electricity bill.”
No published analysis puts an AI-attributable dollar figure on a household bill.Partly, and how much depends on your utility. Estimates of data-center growth’s residential-rate effect range from about -4% (EPRI, 2015-2024) to +2.1% (MIT, 2010-2024); Dominion models a $14 to $37 monthly increase by 2040 for Virginia households. EIA credits most of the 2025 wholesale price rise to natural gas, not data centers.
Open report card“AI caused the transformer shortage.”
The transformer shortage began years before the AI buildout.U.S. distribution-transformer demand is up 41% since 2019 and lead times run one to two years or longer (DOE, 2024), but DOE traces the shortage to post-pandemic demand, aging infrastructure, workforce limits, raw materials and about 40,000 product configurations. Roughly two-thirds of new units go to planned replacements. Data centers add pressure; they did not start it.
Open report card“AI is making ordinary memory and storage more expensive.”
AI memory production is displacing the capacity that makes ordinary DRAM and NAND.Micron reports that producing HBM for AI uses about three times the wafer capacity of equivalent DDR5 (2026), and the BLS storage-device producer price index rose 30.4% between January 2024 and June 2026. The same fabs make consumer DRAM and NAND, so AI demand raises ordinary memory and SSD prices. No public series isolates how many dollars of a given retail price are AI.
Open report card“Every AI prompt uses a bottle of water.”
No per-prompt water figure applies universally, and the bottle fits no current one.It depends on the model, the task, the date and what is counted. Google reports 0.26 mL for a median Gemini text prompt (2025); OpenAI reports about 0.32 mL per ChatGPT query (2025). The bottle came from one modeled GPT-3 scenario: about 500 mL for 10 to 50 responses (2023), not today’s models. A long reasoning or agentic task can use far more than a short answer.
Open report card