Speaking at Y Combinator’s Startup School in San Francisco over the weekend, Huang didn’t mince words about the effect AI will have on the labor market.

“Many tasks will be automated away,” he said in a video published Monday by the famed Silicon Valley startup accelerator. “Every single job will change, and there’ll be a whole bunch of new jobs.”

And yet the Nvidia CEO made clear that tasks are not the same thing as jobs. As such, he pushed back on the narrative of a white-collar bloodbath that some high-profile AI leaders have predicted. Instead, Huang argues, AI will bring more jobs, not fewer.

“The narrative about AI destroying jobs is exactly backwards,” he said. “AI automates tasks away, but it doesn’t necessarily eliminate jobs.”

As companies continue adopting AI, however, the numbers surrounding white-collar displacement look harsh.

Cutting tasks, not jobs

Goldman Sachs said in its AI Adoption Tracker earlier this year that AI was helping wipe out 11,000 net jobs per month in the most AI-affected industries, including marketing, graphic design, and customer service. While that number was an improvement from the roughly 16,000 net jobs being cut per month that Goldman had previously estimated, entry-level and white-collar positions were still the hardest hit.

In the longer term, Goldman senior global economist Joseph Briggs recently estimated, about 9% of the U.S. workforce, or about 15 million people, could be displaced from their jobs as AI spreads over the next decade.

Adding to the gloom and doom are the predictions of AI leaders who foresee mass unemployment, although some have recently softened their tone.

Anthropic CEO Dario Amodei told Axios in May 2025 that AI could eliminate half of entry-level white-collar jobs within one to five years and push unemployment as high as 20%. Amodei has walked this prediction back since then and now says automation may expand human responsibilities. OpenAI CEO Sam Altman, for his part, once said even the job of CEO was not immune to the risk of AI displacement. He also recently backpedaled, saying in May that AI’s rapid development won’t lead to a global “jobs apocalypse.” 

To back up his argument, Huang explained, “The job of a person has a purpose, and that purpose has many tasks.” Even if some of those tasks are automated away, the purpose remains, he argued.

He pointed to radiology, a profession that “Godfather of AI” Geoffrey Hinton famously predicted would soon become obsolete. In 2016, Hinton said hospitals should stop training radiologists because deep learning would outperform them within five years. He later said he spoke too broadly and amended his thoughts in an interview with the New York Times.

Still, contrary to Hinton’s original prediction, radiology hasn’t shrunk. The number of practicing radiologists grew about 12% from 2010 to 2022, according to a study in the peer-reviewed Journal of the American College of Radiology. A separate JACR study projects the workforce will grow another 25.7% to 40.3% by 2055, depending on whether residency positions keep expanding.

“The reason for that is because the backlog of patients is incredibly high,” Huang explained. “Now doctors and hospitals can admit a lot more patients. In order to admit a lot more patients, you need more nurses; more radiologists.”

While AI has increasingly helped radiologists analyze scans, reading images is only part of their job, experts have previously told Fortune. Radiologists also consult with other physicians and monitor patients, while interventional radiologists perform hands-on procedures. If AI takes on the task of reading scans, radiologists turn their attention to other tasks, Christoph Herpfer, an economist and business administration professor at the University of Virginia’s Darden School of Business who studies health care finance and physician labor markets, previously told Fortune. 

The same is true for software engineers, said Huang. Even if AI agents are able to automate the task of writing code, companies will hire more developers to accomplish increasingly ambitious goals.

“The backlog of ideas, the backlog of ambition and aspiration, is so high,” Huang said. “If we can automate away the task of programming, we could hire more software engineers to do more things. We could be more ambitious.” 

To be sure, Huang has previously noted that some pain will come with increasing automation. In January, he admitted some jobs would disappear altogether, as they have during other periods of massive change. Workers who are adept at using AI may also outcompete those who aren’t, he previously said. 

Still, Huang’s broader bet is that companies will use AI’s productivity gains to expand what they produce rather than cutting workers en masse. 

“This is a classic example of productivity increasing growth,” Huang said. “Increasing growth drives more employment.”