Ernie Tedeschi, chief economist at Stripe, finds that while AI tools seem to raise worker-level efficiency in several industries, they are not the primary explanation for the surge in U.S. economy-wide productivity observed in recent years.
The broad picture
Over the past year, output per hour worked in the United States rose by 2.5%, compared with an average annual increase of 1.6% over the previous 20 years, according to Tedeschi. Though the gap may appear modest, if sustained over several years it would compound and significantly raise output and incomes per worker.
What's driving the increase?
Tedeschi's analysis concludes that much of the recent rise in output per person-hour reflects greater utilization of existing capital rather than broad-based AI-driven productivity improvements. Total factor productivity — which accounts for output relative to both labor hours and units of capital — has shown little change.
Looking across industries, sectors with higher AI adoption do exhibit stronger productivity growth, but Tedeschi notes this pattern predates the pandemic and the widespread use of advanced large language models. In other words, the correlation between AI use and higher productivity growth is not clear evidence that recent AI advances are the root cause of the macro trend.
Instead, the boost appears to come from increased capital utilization: longer runs of already-built factories, higher use of paid-for server racks and GPU clusters, and increased occupancy of existing hotel rooms.
Why that matters
Higher capital utilization represents real economic gains, but it differs from micro-level productivity improvements generated by better technology or processes. That distinction matters for assessing whether the recent productivity spike is likely to persist.
This does not rule out the possibility that AI will deliver sizable macroeconomic gains down the road as firms address workflow bottlenecks and integrate new tools more deeply. Tedeschi warns, however, that policymakers and analysts should be sober about AI’s current contribution so they can judge whether the technology is producing a temporary boost or a durable, transformational shift.
Bottom line
According to Stripe's chief economist, the U.S. appears to be in a period of elevated productivity growth and AI is part of the story, but the immediate driver seems to be more intensive use of existing capital rather than widespread AI-driven efficiency gains.
