Column | AI is Making Blue-collar Jobs Great Again, But Not For The Reason People Think

In May, Jensen Huang, Nvidia’s chief executive, addressed the US-based Carnegie Mellon University’s graduating class and described an artificial intelligence (AI)-era job market beyond software and engineering. “AI gives America the opportunity to build again,” he told them, naming electricians, plumbers, ironworkers and builders.

The numbers are substantial. A March 2026 Randstad analysis found the demand for skilled trades up 27 per cent over three years, with construction workers up 30 per cent, welders 25 per cent and electricians 18 per cent. Randstad’s US analysis found trade demand growing roughly three times as fast as demand for desk-based professional roles.

Wages are beginning to follow. In India, Randstad’s analysis found blue-collar wages rising about 5.7 per cent a year, against four per cent for entry-level white-collar roles, while blue-collar job volumes grew by 93 per cent between 2022 and 2026. The comparison is between growth rates, not pay levels, so the gap is narrowing rather than reversing.

AI is doing two jobs in this story, and the headline blurs them. The first is a demand shock: the AI boom is also a construction boom, driven by data centres, transmission lines and power plants. McKinsey & Company estimates that America will need about 130,000 more electricians by 2030, along with roughly 240,000 construction labourers and 150,000 supervisors, to meet data-centre demand. The second is augmentation: AI tools make a technician faster, safer or more accurate on the job. That is the channel the headline implies and it has the weakest evidence.

Jim Farley, the CEO of Ford Motor Company, has been unusually candid. He said it was too early to tell whether AI would help or hurt essential workers, and noted that new technologies have a poor record of making blue-collar workers more productive, automation being his example.

Data points the same way. A Federal Reserve study found that about four per cent of blue-collar workers’ weekly hours involved AI, saving roughly one per cent of their work. For computer and mathematics workers, 11.7 per cent of hours involved AI, saving 2.5 per cent.

Much of the strongest evidence that AI lifts weaker workers comes from office tasks. In one well-known writing experiment, college-educated professionals using ChatGPT became quicker, and lower-skilled participants became both quicker and better. An Indian survey found around 70 per cent of blue-collar respondents say technology helps them at work. That is encouraging, but adoption is not the same as measured productivity.

The sharper claims come from firms that sell the tools. Software company Palantir’s chief technology officer has cited Panasonic Energy shortening a battery-technician apprenticeship from three years to three months. The CEO of a maintenance-software company, writing about asset-heavy industries, attributes a 34 per cent reduction in downtime costs to AI-augmented workflows. Each is one company’s account, not an independent evaluation.

The wage pressure has a plainer cause: too few people. A projection by Deloitte and the Manufacturing Institute puts manufacturing’s requirement at about 3.8 million jobs by 2033, with around 1.9 million potentially unfilled. Retirements compound the problem, with industry commentary describing experienced technicians leaving faster than they can be replaced. Farley has said the intent to build is there, but nothing yet backfills the ambition.

India: The Pipeline Question

India’s supply side is moving. Enrolment in industrial training institutes (ITIs) rose from about 1.25 million in 2022–23 to about 1.47 million in 2025–26, and 570 AI and data labs are being set up in smaller cities. The World Bank has approved an $830-million loan to modernise the ITI network. Tamil Nadu plans to introduce AI and machine-learning trades in 50 government ITIs over five years.

The employability data carry a warning. Employability among ITI candidates rose to about 46 per cent from 41 per cent, yet manufacturers plan just five per cent of hires as freshers, preferring experienced talent. Based on the evidence so far, the boom rewards people who already hold skills and certification. It does not yet open doors for the untrained.

Pressure on the degree route may be pushing the other way. A Stanford study, using automatic data processing, of payroll data found that employment for workers aged between 22 and 25 in the most AI-exposed roles fell roughly six per cent between late-2022 and mid-2025, while older workers in the same roles gained. Apollo Global Management found nothing measurable in blue-collar work, consistent with AI not yet reaching far into physical jobs. If that pattern spreads to Indian services and graduate hiring, trades may start to look like a rational alternative. That is a hypothesis, not a finding.

Trades are cyclical and this boom is tied to a single investment wave. Construction and related trades have had a difficult stretch, and the outlook for skilled workers may be uncomfortably entwined with the technology they help build. The more durable demand lies in maintaining, repairing and operating what gets built, lasting long after the construction peak.

The word implies a golden age. For much of the 20th century, a skilled trade was a respectable route to a middle-class life, and the degree-first consensus of recent decades eroded that. The current boom could restore it if three conditions hold.

Training must keep pace. ITIs need to become places young people should choose, not as the last resort, but with modern curricula, employer links and a credible path to pay. Productivity claims need independent measurement, not announcements by firms selling the tools. And gains must reach workers and not be absorbed by contractors and intermediaries. The last is a hypothesis; evidence on who captures the gains is not yet in.

AI may make blue-collar work great again, but mostly by making the country build more, not by making the work smarter. That is a different bet, and its payoff depends on whether the building, and the maintaining, keeps going.

(Views expressed are personal)

Anubha Singh is Vice-Chancellor, COER University, Roorkee