Recent US productivity gains are solid by post-Global Financial Crisis standards. It’s not a 1990s-style boom, but it’s enough of a revival to make many observers wonder whether we’ve entered a new phase of faster productivity growth, driven by advances in artificial intelligence. Maybe soon. But probably not quite yet.
For starters, labor productivity is a famously noisy and oft-revised statistic. Analysts at the Budget Lab at Yale warn against overinterpreting a few quarters of strong readings, especially when GDP data are still being revised against more current jobs figures. What looks like rising efficiency can also reflect changes in who is working. Example: If lower-wage workers exit the labor force, average output per worker rises even if no one has actually become more productive.
There’s also the timing problem. Much of the AI story so far is investment—companies building data centers, buying software, and upgrading equipment—rather than widespread use that clearly boosts output. In its March 2025 report “A Pulse on Productivity,” the Bank of America Institute describes this buildout as essentially “prep work,” with any productivity gains from actual AI deployment still to come. From the note: “Any additional pickup in productivity due to the actual adoption of AI would be icing on the cake: it could further increase productivity and potential growth.” And this from a recent Goldman Sachs report on AI adoption and its “limited” impact: “We continue to observe large impacts on labor productivity in the limited areas where generative AI has been deployed.
There is, however, a less obviously tech-centric explanation for the recent productivity uptick: a rebound in business formation. After decades of declining dynamism, new-business applications have surged since the pandemic and remain well above pre-2020 levels, according to the new short note “Application Accepted: Business Formation Boom Continues” by John O’Trakoun of the Federal Reserve Bank of Richmond, analyzing US Census Bureau data. Because applications tend to translate into actual firm creation, the recent pickup points to continued startup activity. And importantly, much of that activity is showing up in sectors that have historically added jobs at a faster pace—hinting at a “tailwind for future job creation.”
Added growth, too. Startups function as the economy’s trial-and-error engine: Most don’t last, but the ones that do introduce new ideas, challenge incumbents, and shift workers and capital toward more productive uses.
The relationship between business dynamism and productivity growth is both well understood and underappreciated, at least by non-economists. In the 2024 Aspen Institute analysis “The Recent Rise in US Labor Productivity,” economist Luke Pardue points to the post-pandemic surge in new-business creation as a likely key driver of recent productivity gain. He also notes that earlier declines in startup activity imposed a measurable drag on productivity—suggesting that the recent rebound could provide a meaningful continuing boost if it proves durable.
And by the way, there may be an AI kicker to this dynamism story in how the technology can help entrepreneurs do their thing. “The surge in new US business formation is being fueled by AI and large language models that are dramatically reducing the cost and complexity of launching a company,” says Torsten Slok, chief economist at investment firm Apollo. “As these firms scale, they will create jobs, underscoring that AI is likely to strengthen, not disrupt, the US labor market.”
It’s a pro-growth story about AI that should be given more attention.
The post America’s Productivity Pop Has a Startup Backstory appeared first on American Enterprise Institute – AEI.
