McKinsey’s new projections say artificial intelligence will add 41 million American jobs by 2035 and wipe out 36 million – a net gain of five million, the kind of arithmetic that gets quoted on earnings calls. Read it again and notice what the ledger assumes: that the people who lose the 36 million jobs are the same people who take the new ones. The report answers that question itself, and the answer is that 11 million Americans will need entirely new careers.
The McKinsey Global Institute calls the study Workforce in Motion. The number everyone carried away, reported by Fortune, was the reassuring half – more jobs created than destroyed, a net plus of five million. The number buried a few lines down is the one that matters to the people living inside it: this is not a rounding error in a forecast, it is a forced relocation of an entire workforce, priced in years of retraining that most of the displaced will have to fund, schedule, and survive on their own.
The jobs will exist. Nobody promised they would be in the same town, or pay the same, or belong to the person holding the old one.
McKinsey – A Net Gain of Five Million Jobs Is Not a Net Gain of Five Million People
The report frames the outcome as mobility: skills that travel, pathways into the occupations that grow. As economics, that is correct. As a promise to the person whose job sits on the 36-million side of the ledger, it is a shrug with a footnote. A net figure combines two populations that never actually meet – the people leaving work they spent decades learning, and the vacancies that appear somewhere else, often in a different sector, frequently in a different state.
The coverage of the report is careful to say the new roles are “new AI-era occupations.” That is the tell. The jobs being destroyed are concrete – clerks, analysts, support staff, junior roles whose entire value was that an ordinary person could be trained to do one in a week. The jobs being created are the ones that require the training. Five million net is a statistic. Eleven million career changes is a demolition permit.
The Layoff Ledger – Companies Are Already Cashing a Check the 2035 Chart Promises
Here is the awkward part of forecasting a transition for 2035: the companies have started it early and are calling it efficiency. Trackers that follow announced cuts put U.S. tech layoffs for 2026 in the tens of thousands, with Oracle’s restructuring the largest single event of the year and AI adoption cited as the primary driver across the sector. The cuts arrived while revenue was healthy, which is the detail that should end the argument over whether this is a crisis or a strategy. It is a strategy.
Executives describe the reductions as making room for AI investment. Read that literally, because that is how it works: the payroll becomes the research budget. The shareholder gets a story about the future; the employee gets a number. When the same companies later point at a 2035 chart showing net job creation, they will be describing jobs they have not built, held by people they have not hired, in towns where they may have already closed the office.
You cannot fund the future by firing the present and then take credit for both.
The 11 Million – A Reskilling Voucher Is Not a Career
The transition has a cost and the report is honest that it lands on workers: people need new skills, new credentials, and the ability to move between industries. What no chart prices is the friction. A career change at 52 is not a course; it is a pay cut, a rewritten identity, and a bet that the hiring manager in an unfamiliar field will pick a beginner who is older than the team. The research on what AI-driven layoffs do to the people who stay behind is not flattering to the strategy: job insecurity and the fear of being next quietly destroy the very adoption the strategy depends on. Firms cut staff to buy AI, then wonder why the survivors will not use it.
None of this makes the forecast wrong. It makes the forecast not a plan. A plan would name who pays for the retraining, who guarantees a job at the end of it, and what happens to the person who cannot move. Absent those answers, “workforce mobility” is a polite term for: figure it out yourself, before the severance runs out.
What Changes Next – The Bill Arrives Years Before the Jobs Do
The five-million net gain is scheduled for 2035. The layoffs are happening now, the reskilling is unfunded, and the political machinery that might have smoothed the crossing – wage insurance, portable benefits, real retraining dollars – is not on the table. That mismatch has a schedule, and the schedule favors the cuts.
So watch the wrong indicators, because the headline will keep pointing at 41 million. It is the number that lets everyone feel fine. The honest measure is unglamorous: how many of the 36 million are still employed at comparable pay five years after their job disappears. If the answer is most of them, McKinsey gets to be right. If it is a fraction, we will have bought a labor-market transition with other people’s careers and booked it as a gain.
The math adds up. The people it is written on are still waiting for their turn.
Sources: Fortune, McKinsey Global Institute, layoffhedge, Mirage News.