TL;DR: There is no single global AI talent shortage number, and anyone quoting one is guessing.
What the public data does support: 170 million roles created and 92 million displaced worldwide by 2030, US data scientist employment growing 35% to 2035 against about 4% across all occupations, and a developer population growing fastest well outside the markets most companies recruit in.
The scarcity is concentrated in specialisms, not spread across headcount.
Most articles on this topic open with a precise global figure for unfilled AI roles. No public dataset produces one. Job-opening counts by country, by AI specialism, with supply ratios attached, are not something anyone measures reliably.
What follows is what the evidence actually shows, from named sources you can open, and what it means if you are trying to hire.

Five things that decide AI hiring right now
- The shortage is real in data and security, not across engineering generally.
- Net job creation is positive at 78 million, but the created and displaced roles go to different people.
- Supply is growing fastest outside the markets most companies recruit in.
- Adoption is near universal at 88%. Financial impact is not, at 37%.
- AI fluency on a CV and AI fluency in practice are different things. Test for it.









