AI Skills Are Now the Hardest Hire in the World and Money Is Not Closing the Gap

Employers are raising wages to compete for AI talent, and the roles are still going unfilled. ManpowerGroup's 2026 Talent Shortage Survey, covering more than 39,000 employers across 41 countries, finds AI capability the hardest thing to hire anywhere in the world for the first time, with AI model and application development (20%) and AI literacy (19%) topping the global ranking of scarce skills.

Roughly 72% of those employers still cannot fill their roles. The large companies with the biggest salary budgets are struggling more than the small ones, and the shortage swings from 87% of employers in Slovakia to 48% in China. A company's odds are being set by where it looks for people more than by what it pays them.

The auction nobody wins

Scale is supposed to be an advantage in hiring. Large employers have the compensation bands, the recruiting teams, and the brand recognition that make a candidate return a call. In this market, they are losing anyway. Firms with 1,000 to 4,999 employees report a 75% shortage rate against 64% at companies with fewer than 10 people, an 11-point penalty for being the better-resourced buyer.

Normally, this fixes itself. Big companies decide they want a skill, they outbid everyone for it, salaries climb, more people train for the work, and the squeeze eases. That has not happened here. Wages have gone up, and the jobs are still open, which means the thing standing in the way is not the price.

Jonas Prising, chair and chief executive of ManpowerGroup, points to what employers are screening for. He says companies will have to hire for potential and build AI literacy across their existing workforce. That prescription points at a pool of credentialed candidates too small for everyone bidding on it.

Where you look matters

The same AI skills are far easier to hire in some countries than others. Germany reports an 83% shortage. The US sits at 69%. China, at 48%, is the least constrained major market in the survey.

Scarcity, then, is partly a function of where a company is willing to look. An employer whose search stops at its own metro area inherits that market's rate as a hard ceiling on its hiring. One that can source from Warsaw or Shenzhen is drawing from a pool where employers report far less difficulty. The survey does not test that trade-off directly, though a separate forecast from the recruiter Robert Walters and Native Teams puts US demand for AI professionals more than 800,000 above what the domestic market will supply by 2028, and finds that firms reaching across borders are the ones moving fastest.

Widening the net

Nearly all employers, 91%, are doing something about the problem. Upskilling and reskilling leads at 27%, well ahead of everything else. Schedule flexibility comes next at 20%, wage increases at 19%, and location flexibility and targeting new talent pools sit at 18%.

Four of those five strategies do the same thing, which is enlarge the pool of people who can plausibly take the job. Upskilling creates candidates who did not exist before. Loosening the schedule brings in people who cannot commit to fixed hours. Dropping the location requirement reaches candidates who live too far to commute. Only wage increases compete for the workers who are already available, and that lever sits fourth.

Judgment is the bottleneck

Even in a survey where AI tops the shortage list, the attributes employers say they want most are human. Communication, collaboration, and teamwork lead at 39%, ahead of professionalism and work ethic at 36% and adaptability at 34%. Prising's read is that "AI is not replacing jobs, it is reshaping work", and that the companies best positioned are the ones connecting productivity gains to career growth for people already on staff.

The technology is available to any company that wants it. What stays scarce is the person who can apply it with judgment, and the survey shows those people are not distributed evenly across the map, not reachable at the price large employers are paying, and not going to appear on a résumé with the exact credential a job description asks for. The employers filling these roles are the ones who widened what counts as a qualified candidate and where they were willing to find one.