On a stage in Paris in June, Jeff Bezos delivered the most optimistic line of the AI-and-jobs debate anyone has managed all year. “AI is going to create a labor shortage,” he told the VivaTech audience. “We’re going to have labor scarcity. People are pessimistic because a bunch of smart people are telling them to be pessimistic.” Five months earlier, in January, the company he founded cut roughly 16,000 corporate jobs, and AI was among the reasons it gave.
Both of those things are true at the same time, and holding them in one hand is the whole exercise of understanding the 2026 labor market.
The bull case, stated fairly
Bezos is not saying anything stupid, and it’s worth laying out his argument at full strength before poking at it. His claim is that AI raises productivity, lowers the cost of things, and therefore creates more demand than the current workforce can satisfy. His examples are the good ones. The bulldozer did not put construction workers out of work; it let a crew build things that were previously unbuildable. When ATMs arrived, banks did not fire all the tellers — they opened more branches, because branches got cheaper to run, and ended up employing more people, not fewer. This is Jevons Paradox wearing a suit: make a tool cheaper and people find so many new uses for it that total consumption goes up. Radiologists and software engineers, in this telling, get elevated rather than deleted. Humanity has, as he put it, an endless set of things to invent.
If you squint at the aggregate data, he has a case. US unemployment sat at 4.2% in June. Job openings actually rose. Average hourly earnings kept climbing. A tight labor market is not what mass technological unemployment looks like, and the economists at Gartner project that AI will create more jobs than it destroys starting around 2028.
The number in the other hand
The problem is that “starting around 2028” is doing a lot of work in that sentence, and the people reading this in July 2026 are not standing in 2028. They’re standing in the year where more than 115,000 tech jobs were cut through May, where the outplacement firm Challenger, Gray & Christmas found employers naming AI as the single leading reason for cuts four months running, and where Goldman Sachs pegs the current rate of AI-attributed losses at roughly 16,000 US jobs a month. That last figure is a grim coincidence worth sitting with: Amazon’s January cut was one month’s worth of the entire country’s AI displacement, done by a single company.
And here is the part that makes the Paris speech hard to take entirely at face value. In the same stretch that Bezos was forecasting scarcity, his own CEO Andy Jassy was on an earnings call bragging about the opposite. Jassy described an internal rebuild that “would have taken 40 or 50 people about a year” and was instead finished by “five really smart, AI-forward-thinking people” in 65 days. That is not a labor shortage. That is one team doing the work of ten and the other nine being the point. Amazon is guiding to something like $200 billion in capital spending this year, most of it aimed at building exactly the systems that let five people replace fifty. You cannot fund the machine that thins the workforce and simultaneously promise the workforce will grow, and expect both promises to land on the same worker.
Both can be right, on different clocks
The honest reading is that Bezos and the pessimists are describing the same event at different points on the timeline. In the long run, cheaper intelligence probably does expand what humanity attempts, and probably does pull in more labor to attempt it — the ATM story is real. In the short run, a specific person in a specific seat gets a specific email in January, and no amount of 2028 optimism pays their August rent. The transition Bezos is describing may genuinely have a bright far end. The question that decides whether it’s a triumph or a tragedy is how long the tunnel is, and who’s expected to hold their breath through it.
For anyone reading this from inside a company that just cited AI while cutting, the takeaway isn’t to bet against Bezos — it’s to notice which side of his own contradiction you’re standing on. The roles that survive the tunnel are the elevated ones he keeps naming: oversight, judgment, the specialized and the revenue-attached. The roles that don’t are the task-shaped ones a well-prompted model can eat in 65 days. A labor shortage is coming, maybe. It is not evenly distributed, and it is not arriving on the same schedule as the layoffs.
Sources
- Fortune — ‘AI is going to create a labor shortage’: Jeff Bezos at VivaTech
- Fox Business — Jeff Bezos predicts AI will create a labor shortage, not replace human workers
- The Hill — Jeff Bezos: AI to cause labor shortage, not replace workers
- 24/7 Wall St. — Amazon cut 16,000 jobs while Bezos predicts AI will create a labor shortage
