NVIDIA CEO Jensen Huang does not agree with the so-called “white-collar bloodbath” narrative. In a speech at Y Combinator’s Startup School, Huang responded to various pessimistic predictions about AI. He emphasized the need to distinguish between “tasks” and “jobs”: AI eliminates the former, not the latter.
“Every job will change, and many new jobs will also emerge,” Huang said. “The narrative that AI will destroy jobs gets it exactly backward.”
Huang gave an example of a “task”: Some customer service representatives answer calls, handle complaints, and respond based on information in a database. “That task will be replaced by automation,” he said.
In fact, customer service has long been one of the industries most vulnerable to the impact of AI. Just last week, Uber laid off approximately 10% of its customer service employees, citing the development of AI technology.
However, Huang believes that a task is not the same as a job. “The reason is that a job has a purpose, and that purpose usually involves many different tasks.”
Using radiologists as an example, he said that AI can already automatically perform the “task” of reading medical images. Although AI radiology tools have become highly mature and widely adopted, this does not mean radiologists have stopped reading images. On the contrary, the number of radiology positions continues to grow.
“The reason is that there is a huge backlog of patients waiting for diagnosis and treatment,” Huang said. “Doctors and hospitals can now see more patients.”
Huang also discussed software engineering, an industry profoundly transformed by AI. Today, many programmers have shifted from writing code themselves to directing AI programming tools such as Claude Code and Codex to carry out development work.
In this case, the task being automated is “writing code,” while the number of software engineering positions is instead continuing to increase.
“People have an enormous backlog of ideas, ambitions, and goals,” Huang said. “If we can automate the task of programming, we can hire more software engineers to accomplish more.”
He also mentioned the AI legal technology company Harvey. Huang said that some people had previously worried that Harvey would replace the work of paralegals, but in reality, demand for this profession is “growing at an astonishing rate.”
However, this view does not fully align with official statistics. According to projections from the U.S. Bureau of Labor Statistics (BLS), the number of paralegal positions is expected to “change very little” between 2024 and 2034.
Huang’s assessment of the future labor market stands in sharp contrast to the views of Anthropic CEO Dario Amodei. In 2025, Amodei predicted that AI could eventually eliminate 50% of entry-level white-collar jobs.
Amodei later adjusted his wording, placing greater emphasis on AI’s ability to significantly improve productivity, but he continued to believe that job losses could be an “inevitable” part of AI’s development.
Huang said that AI will indeed enable employees to work faster and more efficiently, which could in turn create more employment opportunities. “This is a classic example of productivity gains driving economic growth,” Huang said. “Economic growth will in turn create more employment opportunities.”