Who Pays for AI?

Resource ledgers

Ten markets, graded by the available evidence

Each ledger explains AI demand’s market effect, who pays or benefits, the supply constraint, the attribution boundary and the evidence that changes the grade.

AI-drivenAI is the main documented cause
AI-contributingAI is one measurable cause among several
Contestedthe available measures do not isolate AI’s effect
Scapegoatedthe evidence does not support the stated blame
SemiconductorsAI-driven

Memory (DRAM & HBM)

3:1HBM-to-DDR5 capacity trade ratio reported by Micron

AI memory demand is squeezing the same production capacity used for ordinary DRAM.

See the evidence β†’
SemiconductorsAI-contributing

Storage (NAND & SSDs)

+30.4%BLS storage-device PPI, Jan 2024 to Jun 2026

Storage prices rose 30.4%; available data does not isolate AI’s share.

See the evidence β†’
Grid equipmentAI-contributing

Transformers & Switchgear

1–2+ yrdistribution-transformer lead time in 2024

Data centers are adding demand to a transformer shortage that started earlier.

See the evidence β†’
Generation equipmentAI-contributing

Gas Turbines

100 GWGE Vernova backlog plus slot reservations, Q1 2026

Data centers occupy turbine slots within a broader power-construction boom.

See the evidence β†’
LaborContested

Electricians & Skilled Trades

757,220U.S. electrician jobs in May 2025

Data-center construction uses scarce trades; national wage data does not isolate an AI effect.

See the evidence β†’
Local buildoutContested

Construction Inputs

No clean seriesnational AI-attributable price measure

Large data-center projects concentrate demand in local construction markets; national data does not isolate the effect.

See the evidence β†’
EnergyAI-contributing

Electricity

11.8%LBNL reference-case share of U.S. electricity in 2030

Data centers are a major new source of demand. Utility tariffs and cost allocation determine the household-bill effect.

See the evidence β†’
Local resourcesContested

Water

>10,000Γ—observed workload-level water-use variation

Data-center water use varies by more than 10,000-fold across locations and system designs.

See the evidence β†’
Grid accessScapegoated

Interconnection Queue

5 yearsmedian request-to-operation time for projects built in 2023

AI did not create the five-year generation queue. Large data-center loads do create new grid work.

See the evidence β†’
FinancingContested

Capital Markets

No measured effectmatched cost-of-capital comparison

AI absorbs huge investment budgets; public evidence does not establish higher borrowing costs elsewhere.

See the evidence β†’