As uncertainty swirls around the global economic order, mounting public debt, and the future of immigration, something remarkable has happened: productivity surged. In the third quarter of 2025, American labor productivity jumped 4.9 percent. A year earlier, it had grown just 1.9 percent; since 2019, the average annual rate has been only 2 percent.
It isn’t just the United States. Even in the U.K., long plagued by low growth, a new estimate puts productivity growth at 3.4 percent over the last six quarters. That could offer relief to countries struggling to support aging populations. It’s too soon to know whether the trend—for which many credit artificial intelligence—will last. But it is encouraging, especially as governments and stock-market investors alike are counting on AI-driven productivity gains to offset years of fiscal profligacy.
Finally, a reason to check your email.
Sign up for our free newsletter today.
In general, productivity compares the inflation-adjusted value of goods and services produced with the inputs required to produce them. Total factor productivity captures gains from capital, labor, and other resources. When output grows without a corresponding rise in inputs, productivity has increased. Most headlines focus on labor productivity—output per hour worked—because sustained gains support faster economic growth and, over time, higher wages.
Still, measuring productivity requires strong, often contested, assumptions. Output is typically measured using GDP, which may not fully capture improvements in consumer welfare or the value of new products that did not previously exist, such as the iPhone. The choice of price index is also controversial, since it may fail to reflect technological improvements or shifts in consumer preferences.
Productivity does not tell the whole story of economic health. Poorer countries often post larger gains simply because they start from a lower base. China’s rapid rise reflected catch-up growth, deploying more capital and labor and absorbing existing technologies. Whether Beijing will remain a major engine of global growth is uncertain, however, especially given signs of misallocated investment.
Comparisons can mislead, too. Europe’s higher output per hour partly reflects fewer hours worked. If Americans worked 20 hours a week, productivity per hour would rise but total output would fall—and the country would be poorer. Longer vacations and shorter workweeks help explain Europe’s lower consumption relative to the United States.
Yet, however measured, productivity is central to prosperity. For most of human history, gains were modest and easily reversed by war or collapse. That began to change after the Renaissance, accelerating with the printing press, which preserved knowledge and enabled innovation to build on innovation. Agricultural breakthroughs freed labor for industry, and technologies like the spinning jenny and steam engine transformed production. From then on, innovation increasingly fed on itself, sustaining growth and steadily improving living standards.
The Industrial Revolution also showed that innovation makes labor more valuable. New capital may replace some tasks but also increases what each worker can produce, raising wages over time—even if the transition is slow and disruptive. Expanding education reinforced this dynamic: universal schooling during industrialization and, later, the rise in college attendance further boosted productivity as new technologies demanded more advanced skills.
The challenge for rich countries, including the United States, is sustaining productivity growth once labor and capital are largely fully deployed. Moving workers from farm to factory can generate dramatic gains, as China did in the 2000s. But once those basic shifts are completed, productivity can stall.
Education, too, brings diminishing returns. America may be confronting that reality now, with a labor market saturated by college graduates. Over the long run, sustained productivity growth depends on new innovations—and on applying them in ways that reshape the economy. The U.S. has proved exceptional in continuing to lead the world in that process.
Yet even in America, productivity has lagged its historical pace, despite an era of impressive, life-altering innovations. Labor productivity growth averaged just 1.5 percent annually between 2004 and 2022, compared with 2.7 percent from 1950 to 1970.
Enter artificial intelligence, which tech leaders hail as a transformative breakthrough. Google CEO Sundar Pichai has likened its importance to harnessing fire or electricity. Others, including Elon Musk and Anthropic’s Dario Amodei, predict large productivity gains but warn of widespread job displacement. Economists tend to be more sanguine, pointing to history: general-purpose technologies typically complement labor, create new occupations, and raise wages over time. Economic historian Joel Mokyr expects AI to follow that pattern. For now, though, it all remains conjecture. AI may rival electricity, or it may prove more like the internet: revolutionary in some ways, but less potent for overall growth than promised.
This is why so much rides on each productivity report. The latest numbers are promising, but it may be too soon to celebrate. AI adoption has surged, though the scale is hard to pin down. Estimates suggest that 20 percent to 68 percent of businesses are using some form of AI. “Adoption” ranges widely, from embedding AI into core workflows to employees using chatbots to draft emails. Still, the spread is unmistakable. And the gains were not a one-quarter anomaly: productivity rose about 2.7 percent in 2024 as well.
Productivity in the nonfarm business sector, where AI adoption has been especially rapid, rose 5 percent in the third quarter of 2025. An August 2024 survey estimated that 28 percent of workers used AI on the job, though usage intensity varied widely by industry. Information services, finance, and professional and business services showed the heaviest use—and have posted notable productivity gains. It remains unclear how much credit AI deserves, or whether such gains will spread across the broader economy. We might expect AI to transform information technology. But will it meaningfully raise productivity in areas like health care?
Erik Brynjolfsson, director of Stanford’s Digital Economy Lab, has described productivity transitions as following a “J-curve.” General-purpose technologies—from the steam engine to the computer—require heavy up-front investment in intangible capital: reorganizing workflows, retraining workers, and developing new business models. During that phase, measured productivity can appear weak as resources are diverted toward investment. Only later does a “harvest” phase show up in the statistics. If recent data persist, he suggests, the American economy may be moving into that second stage with AI—though several more quarters of sustained gains would be needed to confirm a lasting shift.
Still, caution is warranted. By fall 2025, AI adoption had leveled off and remains less common among smaller firms. Productivity growth was weaker in the first two quarters of 2025. In research with Bharat Chandar and Ruyu Chen, Brynjolfsson finds that entry-level hiring in AI-exposed occupations declined by roughly 16 percent, suggesting that firms may be automating some codified junior tasks. At the same time, workers who use AI to augment their skills appear to fare better in the labor market. The gains, in other words, may be uneven, rewarding those who adapt while displacing others.
Other forces may also be contributing to the recent surge. Productivity in durable-goods manufacturing, which is generally less AI-intensive, rose 5.4 percent. Nonfarm business output climbed 5.4 percent, while hours worked increased just 0.5 percent, down from 0.9 percent a year earlier. An aging population and slower immigration are restraining labor supply, mechanically boosting output per worker.
Even if AI proves more consequential than any innovation since the harnessing of fire, its effects may not show up clearly for years. The steam engine took decades to register in productivity statistics. Even electricity ultimately supported long-run GDP per-capita growth of only about 2 percent annually. To meet today’s debt burdens and entitlement promises, we would need faster gains than that.
Nor do we know what the technology will mean for wages. Some AI-complementary jobs may see pay rise; others may disappear. As in earlier transitions, new occupations will take time to emerge. Encouragingly, white-collar employment and wages have grown in recent years, despite predictions that these roles would be first to go.
What we do know is that productivity is central to our economic future. It may come from AI, from technologies that AI helps unlock, or from something else entirely. But we cannot passively wait for rescue. Sustained growth requires pro-innovation policies that keep the economy dynamic and adaptable: lighter regulation, stronger incentives for research and development, and immigration reforms that attract and retain skilled, entrepreneurial talent. That is the surest way to raise productivity—and with it, long-term prosperity.
