Electricians wanted, graduates not: the AI jobs split behind the midterm mood

The class of 2026 left campus into a job market that is weakest at the bottom rung. Unemployment among recent college graduates has stayed above the overall rate for five consecutive years, according to New York Fed data cited by JBizNews, and was near 5.7% in the first quarter, when the national rate stood around 4.3%. Meanwhile, builders of AI data centers are being warned about thin benches of skilled trades workers, with electricians among the occupations Chmura Economics expects to be stretched at peak construction.

That split sits under the AI anxiety that came through in Tuesday's midterms. The evidence that AI is already causing broad job losses is still contested, though, and what employers say about AI has drifted a long way from what the employment numbers show. With some races still being counted, the fair summary is that the anxiety voted. The jobs data has not settled the question of what caused it.

The worry was real. The cause is unproven

Before the vote, AI registered as a real concern that still ranked well below the economy. In Pew Research Center's survey fielded September 28 to October 4, 84% of registered voters called the economy very important to their vote, and 44% said the same of artificial intelligence. That 44% splits by party: 54% of Democratic-candidate supporters versus 34% of Republican-candidate supporters. An NBC News exit poll reported by TIME, in which 56% of voters called AI and data centers a crucial or major factor in their vote for Congress, bundles AI with local fights over power and land and measures how much voters cared, not what changed their minds. Nothing in the exit data shows that AI decided any particular race.

Neither party has won voters' trust on the issue. In an NBC News poll from March, reported by Fortune, 20% of Americans said Republicans would do a better job governing AI and 19% said Democrats would. Ali Mortell, director of research at the progressive firm Blue Rose Research, told TIME that this was the first cycle in which AI showed up in political debate, and said its effects on jobs and safety could make for a very different argument by 2028.

What employers say: a layoff ledger that peaked in May

The figure most often cited for AI job losses comes from Challenger, Gray & Christmas, which counts announced US job cuts and the reasons employers give for them. Through September, employers cited AI for 120,136 cuts, about 21% of the 573,195 total. That is more than double the 54,836 AI-attributed cuts Challenger counted for all of 2025. AI was the top stated reason for five straight months from March through July. It peaked in May, when 38,579 cuts, roughly 40% of that month's total, were blamed on AI. AI ranked fourth in August with 3,462 cuts and fifth in September with 3,961. As of Wednesday morning, Challenger's site still listed the September report as its latest, so the October numbers were not yet available.

The count has limits. It tallies announcements in which a company named AI as a reason. It does not show that software actually took over the work. Total announced cuts through September were 39% lower than the 946,426 Challenger counted in the first nine months of 2025, which is hard to reconcile with a story of mass displacement. In April, Andy Challenger said that even where AI is not replacing individual jobs, it is claiming the money that would have paid for them. In July he said AI was shifting the labor market without dismantling it.

Some economists suspect part of it is cover. Martha Gimbel, executive director of the Yale Budget Lab, has linked some AI-attributed layoffs to companies' difficulties with tariffs, immigration limits and wider policy uncertainty, Fortune reported, and has pointed to post-pandemic hiring and higher interest rates as bigger drivers of today's low-hire labor market.

What the data shows: no economy-wide shock, but fewer entry-level hires

A May analysis by the Yale Budget Lab, titled "AI Is Probably Not (Yet) the Reason for Labor Market Weakening," estimated that the effect on employment in AI-exposed occupations was close to zero and statistically indistinguishable from it, with a similar result for wages. Unemployment did rise, from a post-pandemic low of 3.4% in April 2023 to 4.3% in March 2026, but the Budget Lab found little sign that AI exposure explains it.

The strongest evidence of harm concerns young workers. In an August update, Stanford Digital Economy Lab researchers Erik Brynjolfsson, Bharat Chandar and Ruyu Chen estimated that employment of 22- to 25-year-olds in the most AI-exposed occupations was about 19% below where it would be if it had kept pace with less-exposed peers as of June 2026, up from 15% in the previous year's data. Most of it reflects fewer hires of young people, not more firings. The gap holds up after controlling for exposure to interest-rate increases and to remote work, and it shows up mainly in jobs where AI automates tasks rather than assisting the worker. The gaps shrink when education is accounted for, and some of the divergence predates generative AI. The authors are explicit that these patterns are descriptive and that they cannot yet say whether AI is the cause.

The New York Fed's latest numbers add a complication. Unemployment among recent graduates fell to 5.2% in the third quarter, its lowest in more than a year, while underemployment rose slightly to 42.5%. More graduates are finding work, but more of that work does not require a degree. The bottom rung is narrowing. It has not disappeared.

The Trump administration's view is that AI leadership will produce jobs. A White House official, speaking anonymously to Bloomberg in reporting published by Fortune last December, argued that AI would aid rather than replace American workers and would bring lower costs for goods and services. National Economic Council director Kevin Hassett has said AI will make productive workers a great deal of money.

Hard hats and hiring freezes: labor splits over the buildout

Chmura Economics puts direct peak construction employment at roughly 0.7 to 2.0 workers per megawatt in general, though its own model of a 64 MW hyperscale facility projected far more: about 300 workers on average and around 750 on site at peak over a 20-month build. A guide from the United Association, the pipe trades union, estimates that 280,000 to 850,000 of its members could be needed if planned projects go ahead at the same time. The same guide says a finished medium-size facility employs only 15 to 80 people directly for the operator, though maintenance and service contracts often add more.

Organized labor has split along that line. North America's Building Trades Unions, which represents more than 3 million workers, has partnered with OpenAI and Microsoft on construction and training. The IBEW, the electricians' union, urged members to lobby against data-center moratoriums, Bloomberg Law reported. On the other side, National Nurses United and the American Association of University Professors endorsed Sen. Bernie Sanders' proposed moratorium. Steamfitters Local 602, which has about 6,100 members in Maryland, Virginia and D.C., told candidates that opposing data centers would cost them the union's backing. "We are dependent on these jobs," business manager Sidney Bonilla told The Daily Wire. The AFL-CIO's convention tried to straddle the divide; it called for union-built data centers, for tech firms to pay their share of energy and water costs, and for workplace AI guardrails. President Liz Shuler called the either-or framing "a false choice."

Jason Shedlock, president of the Maine Building & Construction Trades Council, put the trade-off plainly to E&E News during a fight over a Maine data-center bill: "The construction industry is about a string of temporary jobs." The jobs AI is creating are physical, temporary and local. The jobs it may be squeezing are white-collar, permanent and entry-level. They go to different people in different places, so a construction boom in one county does nothing for a laid-off analyst or an unhired graduate in another.

Rising electricity bills, which many voters link to the buildout, added to the pocketbook mood.

What a new Congress could actually do

The proposal with the clearest bipartisan backing is about measurement. Sens. Josh Hawley, a Missouri Republican, and Mark Warner, a Virginia Democrat, announced the AI-Related Job Impacts Clarity Act a year ago. It would require large companies and federal agencies to report AI-related layoffs and displacement to the Labor Department every quarter, with the results made public. "Artificial intelligence is already replacing American workers," Hawley said at the time. Until something like it exists, the most-cited national figure comes from corporate press releases.

The redistribution proposals go much further. Sen. Mark Kelly's Make AI Work for Americans Act, unveiled this fall, would fund support for workers through taxes on excess profits, digital advertising and AI usage. New York's Alex Bores campaigned on an "AI Dividend" that would pay displaced workers once federal indicators showed AI meaningfully displacing jobs. He proposed funding it through equity stakes in AI firms or a per-token tax on AI use. Bores lost his primary despite $21.3 million in pro-safety outside support, though the winner, Micah Lasher, had co-sponsored the same AI safety law that made Bores an industry target.

Several data points will test the anxiety against the evidence over the next few months: Challenger's October report, the next vintage of the Stanford data, and whether job-loss reporting gets a hearing in the new Congress. Many voters already see AI as a threat to their livelihoods. Economists have not yet shown that it is, or settled how large the effect is.