The numbers come first. The share of tech job postings requiring explicit AI fluency hit 75% in June, up from 67% in March and a 178% jump from a year ago, according to data cited by economist Ed Yardeni. That single figure captures the speed at which the labor market is rewarding one kind of worker and quietly retiring another.
Entry-level corporate jobs are drying up. Whether driven by AI automation or the lasting structural effects of remote work, the traditional on-ramp for young professionals is narrowing. At the same time, demand for skilled tradespeople is booming as companies race to build the physical infrastructure that powers AI — data centers, power lines, cooling systems. Jensen Huang has publicly noted that 'a lot' of six-figure jobs in plumbing and construction will soon open because someone has to build those facilities.
The demographic math makes the mismatch worse. Baby boomers are retiring en masse, and decades of declining birth rates have left a thin pipeline of replacement workers. Lightcast, a labor market data company, predicted in 2024 that U.S. employers will face 'the largest labor shortage the country has ever seen.' Georgetown University's Center on Education and the Workforce estimated the U.S. economy will need 5.25 million more workers with post-secondary education and training, warning that 171 occupations face skills shortages through 2032 without major increases in credentialing.
AI is also reshaping who starts a business. Yardeni pointed to a surge in openings at companies with just one to nine employees, arguing that AI tools have lowered the cost and complexity of entrepreneurship enough to accelerate business formation. Yet actual hiring remained muted even as openings rose — a signal, he wrote in a June note, that 'the job openings AI creates are highly specialized, so talent is in short supply.' He invoked the Jevons Paradox: as AI makes tasks more efficient, total demand for that capability ultimately rises, making AI a net creator of jobs over time.
Nobel laureate Simon Johnson, an MIT economist, frames the choice differently. In a June interview with the Financial Times, he described two paths using his own daughters as examples. One is drawn to lab science — high-judgment, human-intensive work that AI cannot easily replicate. The other is a generalist, and Johnson argues the moment calls for what he terms a 'general-purpose nerd': someone who can master a new AI tool over a weekend, conduct face-to-face interviews on Monday, parse a 50-year-old undigitized book on Wednesday, run a podcast on Thursday, and draft a one-page executive memo on Friday. That profile, he said, is 'quite appealing' to leaders. A research assistant who merely retrieves stock data, by contrast, is easily replaced.
The fastest-growing skill sets, according to tech workplace platform Dice, are in agentic AI, responsible AI, and AI infrastructure — all highly specialized and all commanding a premium.
---
The market has already voted. Capital and hiring managers are paying up for workers who either go deep into AI-adjacent specializations or who combine genuine human judgment with the agility to adopt new tools fast. The generalist who does neither — the corporate middle-layer that once absorbed millions of college graduates — is the profile most exposed.
For free-enterprise advocates, the lesson is straightforward: the workforce that thrives will be one shaped by individual investment in skills, not by credential inflation or bureaucratic retraining programs. The entrepreneurs multiplying at the one-to-nine-employee tier are proof that when the cost of starting a business falls, Americans build things. Policy that protects incumbents and raises the regulatory cost of hiring will only deepen the mismatch the data already shows.
