Erik Brynjolfsson has spent the last several years building one of the most detailed empirical pictures of how technology is reshaping the American workforce—and the picture keeps getting darker for workers at the bottom of the corporate ladder.
Last August, the Stanford economist, who has been a thought leader on artificial intelligence (AI) for years, made headlines when he and his team published a first-of-its-kind study revealing the AI revolution was already having a “significant and disproportionate impact on entry-level workers in the U.S. labor market,” particularly young people ages 22 to 25 in white-collar fields like software engineering and customer service.
Now, in a new working paper published through the National Bureau of Economic Research this February, Brynjolfsson and a team of co-authors have trained their lens on blue-collar America—and found minimum wage increases are accelerating the adoption of industrial robots on factory floors.
Taken together, the two papers trace the outlines of a labor market transformation that is squeezing workers from both ends: AI encroaching from the top, automation moving in from the bottom.
The white-collar warning shot
The August 2025 study was built on an unusually powerful dataset—high-frequency payroll records from millions of American workers generated by ADP, the largest payroll software firm in the country. What Brynjolfsson and his co-authors found was striking: Since the widespread adoption of generative AI tools beginning in late 2022, employment for early-career workers in the most AI-exposed occupations fell by 13% on a relative basis, even after controlling for broader firm-level disruptions. Older, more experienced workers in the same fields, meanwhile, saw their employment hold steady or grow.
The new study, co-authored with J. Frank Li of the University of British Columbia, Javier Miranda of Germany’s Halle Institute for Economic Research, Robert Seamans of NYU’s Stern School of Business, and Andrew J. Wang of Stanford, turns from algorithm to assembly line. Using confidential U.S. Census Bureau microdata linked to customs import records, the team tracked industrial robot adoption among roughly 240,000 single-unit U.S. manufacturing firms from 1992 to 2021—identifying robot adopters by the moment they began importing machines from overseas suppliers in Japan, Germany, and Switzerland.
The central finding is precise and consistent: A 10% increase in the minimum wage is associated with an approximately 8% increase in the likelihood a manufacturing firm will adopt industrial robots, relative to the average adoption rate in the sample.
“Firms subject to higher minimum wages are more likely to adopt robots,” the authors wrote, “even after controlling for observable firm and local economic characteristics.”
The logic mirrors the white-collar story, even if the mechanism is different, with the authors arguing these effects are “economically meaningful.” Just as AI becomes economically attractive when it can replace the codified work of a junior software engineer or customer service rep, an industrial robot becomes more attractive when the cost of the human doing repetitive assembly or welding goes up. In both cases, a rising price for labor at the lower end of the skill spectrum tilts the calculus toward machines.
“While robots may enhance productivity,” Brynjolfsson and his authors wrote, “they may also alter the structure of employment, especially in low-wage sectors as typically found in manufacturing.”
A rigorous test
The manufacturing study’s most compelling evidence comes from a geographic quasi-experiment. Rather than simply comparing firms in high-wage states to those in low-wage states—an approach vulnerable to the objection that those states differ in countless other ways—the researchers focused specifically on companies located in counties that sit directly on state borders, comparing businesses on opposite sides of the same line. These firms face nearly identical local economies, labor markets, and industries. The only meaningful difference is which state’s minimum wage law applies to them.
Under this stringent border-pair test, a 10% minimum wage increase was still associated with an 8.4% rise in robot adoption—a figure that held up across multiple regression specifications and closely matched the broader aggregate analysis the team conducted at the state level. The effect was robust to controls for firm size, age, industry, and whether a state had right-to-work laws on the books.
A pattern across borders
The finding is not unique to the U.S. A study of Turkey found a sharp 33.5% minimum wage hike in 2016 drove medium and large firms to increase robot use, particularly in industries heavy with blue-collar, routine-task workers.
Research in China found similar dynamics from 2008 to 2012, with a 10% minimum wage increase raising the probability of robot adoption, with stronger effects at high-productivity and private-sector firms.
German researchers examining the country’s minimum wage introduction in 2015 found plants with high shares of simple manual workers in routine tasks were the most likely to respond by adopting robots.
The policy tension
Brynjolfsson and his co-authors were measured in their conclusions, appropriately for a non-peer reviewed working paper. The manufacturing paper does not attempt to measure downstream employment effects—whether workers displaced by robots find new jobs, or at what wages—and the authors acknowledge robot adoption can sometimes correlate with higher firm-level productivity and even employment growth, as some international firm-level research has found.
But on the central policy question—whether minimum wage increases drive automation—the evidence is now hard to dismiss. And given Brynjolfsson’s August finding AI is simultaneously eroding the entry-level white-collar labor market, policymakers face a compounding challenge: two distinct technologies, encroaching on two distinct segments of the workforce, through two distinct mechanisms, at the same time.
“Policymakers may wish to consider complementary strategies to mitigate potential displacement effects,” the authors wrote, “such as retraining programs or targeted support for small firms” a prescription that, in light of the parallel AI findings, may be arriving in timely fashion.
For this story, Fortune journalists used generative AI as a research tool. An editor verified the accuracy of the information before publishing.
