AI job cuts slow in August as entry-level hiring gap widens, new data shows

That single month does not settle the argument. For the year through August, AI remains the most-cited reason for announced cuts. The same dataset shows 529,914 total announced cuts, 41% lower than the same period in 2025. AI is a bigger share of a smaller number, which is a different problem than the one most headlines describe.

The timing matters because the labor market debate has moved from whether AI destroys jobs to where in the workforce the damage lands. Newer research suggests the answer is narrower and more specific than the aggregate numbers imply: younger workers, entry-level roles, and a handful of fields.

Challenger's method is worth stating plainly. It records the reason an employer gives when it announces cuts. Its July report flagged that the line is blurry. Visa's July cuts were counted as AI cuts because AI played a significant role, though not the only one. That is a judgment call, not a measurement, and it should temper any reading of the series as a precise scoreboard.

Blue-collar demand complicates the story

While white-collar AI displacement dominates the coverage, CNBC reported on 26 September that the buildout itself is creating trade jobs at a pace that surprises economists. Welders, plumbers, HVAC technicians and electricians are in high demand for data center construction, though many of those roles are temporary by nature.

Nicole Bachaud, a labor economist at ZipRecruiter, told CNBC that the mean minimum salary for data center jobs spiked 125.1% year over year to nearly $208,000. She attributed the figure to highly specialized engineering roles pulling the average up. Postings for welders and pipefitters are up 164% year over year, according to Bachaud's data.

Maria Flynn, president and CEO of the nonprofit Jobs for the Future, told CNBC that data centers are only one piece of a larger puzzle requiring grid modernization and transportation upgrades. "We are seeing major sources of demand that are converging at the same time," she said.

"These projects create demand for on-site workers and therefore cannot be offshored or conducted remotely, unlike manufacturing and software."

That quote comes from Justin Sinkovich of Columbia College Chicago, also speaking to CNBC. He pointed to Louisiana, where Amazon committed $12 billion to a data center with 540 on-site jobs, including 1,700 electricians, technicians and security personnel. Meta's Hyperion project is valued at $27 billion. The geographic pattern is uneven: Bachaud told CNBC that Houston and Birmingham have seen particularly strong growth, while coastal hubs face higher costs and more regulatory burden.

CNBC also noted that recent moves by states including New York and Texas to slow data center development, plus local freezes in response to public backlash, could hit that labor trend. The blue-collar boom, in other words, is politically contingent.

The entry-level gap is where the data converges

The most consistent finding across the newest research concerns workers aged 22 to 25. Stanford's Digital Economy Lab, tracking millions of workers through ADP payroll records, found that employment for that group in AI-exposed jobs sits 19% below trend as of the August 2026 revision of the paper. This is up from 13% on data to July 2025 and 16% to September 2025.

The authors, Brynjolfsson, Chandar and Chen, report that the effect works through fewer hires rather than firings. It is concentrated in jobs where AI automates work instead of assisting it. They also state there is "no evidence of widespread, economy-wide job displacement."

A separate firm-level study summarized by jobsdata.ai complicates that picture. Chandar and Klein Teeselink instrumented AI adoption off job ads involving generative AI and traced outcomes inside foreign affiliates of adopting companies against matched controls. The study covered 1.25 billion job postings and 154 million employment records across 41 countries. They found the junior share of employment at adopters fell 1.9 percentage points by March 2026, but mostly because senior employment rose 6.7%, not because junior employment fell. The junior change was minus 2.5% and not statistically significant, and total employment rose 3.3%.

Their reading is that AI is labor expanding rather than labor saving, with the cost landing on who gets hired rather than on how many are employed. Among technology affiliates the pattern is sharper: junior share down 3.9 points, senior employment up 14.6%, total up 7.8%. Two caveats apply, and jobsdata.ai states them: the design answers what happens inside adopting companies, not to a national labor market. Its definition of junior refers to seniority, not age 22-25, so it is not the same quantity as the Stanford series.

Majors, not occupations

A third study, summarized by jobsdata.ai, identifies AI exposure by field of study rather than occupation. This sidesteps the problem that employers may pull back from roles they expect AI to take rather than work it already does. Using Census PSEO and LEHD administrative records covering 6,665,500 bachelor's graduates, about 29% of all US bachelor's degrees conferred between 2016 and 2024, it found graduates in the most AI-exposed decile of majors became 5 percentage points less likely to be employed in the quarter after graduation and earned about 13% less.

Both effects started immediately after ChatGPT's release. The recession literature puts initial earnings losses from graduating into a downturn at 9-10%, so this is worse, though concentrated in a few fields rather than economy-wide. Roughly half the decline came from lower pay inside the same industries and half from graduates moving into worse-paying ones. The share entering Professional, Scientific and Technical services fell almost 6 points and Information over 3, while Accommodation and Food Services and Retail each gained more than 2 points. Counting that shift, top-decile earnings fell 15%, to levels last seen before 2016. The effect attenuates to about 5% after two years, and the sampled institutions skew large, public and research-heavy.

Second Talent's 24 September roundup puts the aggregate in payroll terms. Goldman Sachs economists estimate AI reduced US monthly payroll growth by roughly 16,000 jobs over the past year and raised unemployment by 0.1 percentage point. Roles where AI assists workers added about 9,000 jobs a month. Goldman adds that the true net effect is probably smaller, because the estimate leaves out data center construction and demand created by higher productivity.

Second Talent also reports that US employers named AI in 116,175 announced job cuts from January to August, about 22% of the total. It also notes that 26.1 million US workers are both highly exposed to AI and poorly placed to adapt, with 86% of them women. It cites PwC's 2026 count showing workers with AI skills earn a 62% wage premium, and a New York Fed finding that recent college graduates had an unemployment rate of about 5.6% in the second quarter of 2026, against 4.1% for all workers.

The sources disagree on scale rather than direction. Challenger's cut announcements, Stanford's payroll gap, Goldman's payroll estimate and the firm-level dilution finding all point to pressure on entry-level hiring. What they do not support is the claim that AI is producing mass unemployment. Yale's Budget Lab finds no clear sign of AI disruption across the labor market as a whole, according to Second Talent. The New York Fed's September surveys found 61% of the region's service firms used AI while just 4% had laid anyone off because of it.

For now, the numbers describe a labor market being reshaped at the point of entry, not one being emptied out. The blue-collar buildout is real and dated. So is the entry-level gap. Both can be true, and the policy fight is already forming around which one gets treated as the headline.