Jensen Huang and Bill Gates are now arguing in public about the only question that matters to everyone else: whether the AI they are building will leave anyone a job. The answer, sitting in the Federal Reserve’s own numbers this week, is that both men are right — and being right is exactly what lets them both off the hook. The jobs Huang is promising and the jobs Gates is mourning are not the same jobs. The graduate who cannot get hired this fall and the electrician who cannot be hired fast enough are living in two different economies, and the billionaires are each describing only the one that flatters them.
Here is what actually happened this week, stripped of the press-release gloss. Nvidia’s chief executive went on Fox Business and promised that AI will create hundreds of thousands of jobs — chip plants, packaging lines, computer factories, the entire physical build-out he calls AI factories. He was answering Bill Gates, who had just warned in a letter that the technology is being underestimated and could produce mass unemployment. Two of the richest men on the planet, both now past sixty, selling ordinary workers two opposite futures with the same straight face. Fortune has the exchange. Somebody’s livelihood depends on which one you believe, and neither of them has to live with the answer.
Jensen Huang — The Jobs He Promises Aren’t the Jobs AI Is Erasing
Huang is not wrong about the build-out; the numbers behind it are genuinely enormous. Nvidia alone committed up to $105 billion toward an OpenAI data center in Piketon, Ohio, and the company running it is hiring a power-generation project manager at $140,000 to $170,000 a year — more than double the median household income of a county where fewer than 27,000 people live, as Fortune reports. Skilled trades are undersupplied by an estimated 2.1 million workers nationwide by 2030. When Huang says the people who build things with their hands are about to have a decade, he is describing something real. A man worth $180 billion can afford to be generous about the future, because the future he is describing is the one he is paid to build.
Bill Gates — He’s Right About the Flood and Wrong About Who Drowns
Gates’s warning is also real, and the Federal Reserve has spent three years quietly proving it in a way neither billionaire has bothered to mention. The New York Fed’s August survey found 61 percent of service firms now use AI, up from 25 percent two years ago, and 51 percent of manufacturers — double last year. But here is the part the headlines skip: only 4 percent of service firms actually laid anyone off, and no manufacturer did at all. The median firm has a mere 17 percent of its workers touching AI. The catastrophe Gates warns about is not showing up as layoffs. It is showing up as something quieter and more cruel: a hiring pullback aimed at the people who have never held the job yet. The flood is real. It just does not drown the people already inside the building.
Samuel Dodini — The Dallas Fed Just Found Out Who Loses First
The sharpest evidence this week comes from two economists at the Dallas Fed who matched millions of real job postings against a measure of which tasks AI can actually perform. After ChatGPT arrived in late 2022, postings for the most automatable occupations fell 5 percent by the end of 2023 and roughly 8 percent by early 2025 — and the pullback came from surviving, established firms, not failed startups. The most exposed roles are not factory workers. They are software developers, web designers, clerical workers, editors. A medical records technician now finds nearly 5 percent of her daily tasks handled by a model. And because fewer than half of these posted jobs demand more than two years of experience, the decline lands hardest on the one group that cannot absorb it: recent college graduates. The Fed’s own words, reported by the Dallas Fed, are that the shift negatively affected the labor-market outcomes of recent college graduates from Texas universities and even pushed current students to change what they study. The machine did not fire anyone. It just stopped printing the on-ramp.
The Texas Graduate — The On-Ramp Is Being Dismantled, Not the Factory Floor
Put the two data sets side by side and the story resolves into a single line. The jobs being created and the jobs being erased belong to different people, and the tech industry is in no hurry to point that out. Huang needs electricians in Ohio; the AI never touched the electrician’s job. Gates worries about the clerical worker and the junior coder, and the AI is quietly automating exactly those first rungs — the grunt work that used to be how a 22-year-old got a foot inside a career. You cannot retrain your way from a vanished entry-level coding job to a $160,000 power-project manager in Piketon overnight, and nobody is pretending the person losing the first job is the same person handed the second. The Federal Reserve’s full survey, which you can read at the New York Fed’s Liberty Street Economics, keeps repeating the same word — retraining — as if a word were a bridge.
There is a comfortable version of this story where the labor market is simply evolving, where the electrician’s boom and the graduate’s freeze are two halves of one healthy adjustment. That version is comfortable, and it is also a lie by omission. The layoffs have not come, but the on-ramp is being dismantled one job posting at a time, and nobody with power is measuring the cost in the only currency that matters: the specific graduate in Texas who did everything she was told to do, earned the degree, and now finds the first job she was supposed to qualify for quietly handed to a model that never asked for a salary. Huang is busy building the exits. Gates is busy warning about the door. Somewhere in between stands a person, and neither of them is paying her any attention. The genie did not just escape the bottle. It redesigned the ladder.
Sources: Fortune, Dallas Fed, New York Fed (Liberty Street Economics).