Cleveland Fed (one of the Federal Reserve’s 12 regional banks) released a
model-based paper finding permanent AI price cuts produce larger long-run output
gains than merely increasing AI usage share. In the baseline model, doubling AI
usage from ~2.4% to 4.8% raises AI capital stock ~81% but, because existing AI
capital remains scarce, creates a bottleneck that lowers long-run output by
about 1.9%. A permanent 50% decline in AI prices increases effective AI capital
~89% and raises long-run output by roughly 1%. Simultaneous price declines and
greater usage are mutually reinforcing—price cuts relieve capital scarcity while
higher usage boosts the marginal output of new AI capital—but in the paper’s
baseline the combined output effect is still smaller than the price cut alone.
The authors define AI capital as firms’ durable AI-related hardware stock,
modeled as a distinct production factor akin to plant and equipment.
