Technology giants including Microsoft Corporation (MSFT  ), Meta Platforms, Inc. (META  ), Amazon.com, Inc. (AMZN  ) and Alphabet Inc. (GOOGL  ) (GOOG  ) have committed hundreds of billions of dollars toward AI infrastructure, while investors continue to question whether those investments will ultimately generate adequate returns.

Nvidia Corp (NVDA  ) CEO Jensen Huang sees a different reality emerging.

Speaking on Nvidia's latest earnings call, Huang argued that artificial intelligence has reached a turning point where generating AI output is no longer just a cost center-it is becoming a profitable business.

"The reason is simple. Agentic AI has arrived. AI can now do productive and valuable work. Tokens are now profitable," Huang said. He added that model developers are now racing to expand capacity as AI-generated output increasingly translates into revenue.

Why Profitable Tokens Matter

In simple terms, a token is a unit of AI-generated text, code or content.

For years, the concern was that serving AI models would remain too expensive, limiting adoption and weighing on profitability. Nvidia argues the opposite is happening.

Earlier in the call, Nvidia CFO Colette Kress highlighted how the company's latest systems continue to reduce the cost of AI inference. Kress said Nvidia's GB300 platform delivers a 60% reduction in cost per token compared with systems available just six months ago.

Lower costs, combined with rising demand for AI services, could significantly improve the economics for model developers and cloud providers.

Nvidia Bull Case

Huang's remarks may also explain why Nvidia remains confident that AI infrastructure spending is still in its early innings.

If AI-generated output is becoming profitable, demand for computing power may continue to rise rather than slow. More profitable AI applications would likely require more inference capacity, more data centers and ultimately more chips.

That logic underpins Nvidia's broader view that the build-out of AI factories is accelerating, and that AI infrastructure spending could eventually reach trillions of dollars annually.

For investors, Huang's "tokens are now profitable" remark may have been one of the most revealing comments of the earnings call-not because it described Nvidia's latest quarter, but because it offered a glimpse into why the company believes the AI spending boom is far from over.