AI hiring crunch: numbers

The hiring market for artificial intelligence now has a strange shape: companies are posting far more roles, but the hardest part is not getting budget approved, it is finding people who can actually do the work. ManpowerGroup’s 2026 survey says 72% of employers worldwide are struggling to fill jobs, and artificial intelligence skills have moved to the top of the shortage list for the first time. (manpowergroup.com) That shortage is not just about “artificial intelligence” in the abstract. ManpowerGroup says the single hardest skills to find are artificial intelligence model and application development at 20% and artificial intelligence literacy at 19%, ahead of traditional information technology and data skills, which fell to seventh place at 17%. (manpowergroup.com) Manatal’s February 27, 2026 recruitment report shows why recruiters feel the squeeze. Its data says artificial-intelligence-focused job postings rose 3.1 times from 2023 to 2025, and 69% of those openings were concentrated in machine learning and data science rather than in broad “use artificial intelligence somehow” jobs. (manatal.com) That split matters because machine learning and data science are the engine room jobs. They are the roles that build models, tune systems, clean training data, and turn a demo into a product that can survive real customers and real error rates. (manatal.com) The market has been moving this way for years, not just since chatbots went mainstream. PwC’s 2024 Global Artificial Intelligence Jobs Barometer, based on more than half a billion job ads across 15 countries, found postings requiring artificial intelligence skills were growing 3.5 times faster than all job postings, and jobs needing those skills carried wage premiums of up to 25% in some markets. (pwc.com) The demand is also spreading beyond the companies people usually think of as “tech.” CBRE found non-tech industries hired more tech talent workers than the tech industry in 2023, and artificial intelligence’s share of total United States tech talent job postings rose to 14.3% in June 2024 from 8.8% in late 2019. (cbre.com) That helps explain why the shortage feels broader than Silicon Valley. A bank adding fraud models, a hospital buying clinical software, and a manufacturer automating quality checks can all end up chasing the same machine learning engineer at the same time. (cbre.com) Even inside the United States, the pressure is uneven. ManpowerGroup says 69% of United States employers report hiring difficulty, which is slightly below the 72% global average, while Germany is at 83%, France at 74%, and the United Kingdom at 73%, so companies are competing in a global market for a talent pool that is still limited. (manpowergroup.com) The quiet twist in the data is that companies are not only short of specialists. They are also short of people with enough artificial intelligence literacy to use the tools well, judge outputs, and fit them into normal workflows, which is why “artificial intelligence literacy” nearly ties “model and application development” in ManpowerGroup’s ranking. (manpowergroup.com) So the bottleneck is now two layers deep: too few people can build the systems, and too few people outside the engineering team can work confidently with them. That is why job postings can surge at the same time employers keep saying they cannot hire, and why the artificial intelligence labor market still looks tight even after the first wave of hype. (manatal.com) (manpowergroup.com)