The company has profited handsomely from an AI craze that has seen tech giants pour billions of dollars into Nvidia chips to help power large language models like ChatGPT and Gemini. But doubts remain as to whether these companies are actually making money on AI, and whether they can continue their frenzied pace of spending. 

When Huang was asked on the company’s earnings call Wednesday about the sustainability of customer spending, he raved about AI’s ability to transform computing, but danced around the question.

For eight minutes, Huang talked about how the usual CPUs that are standard the world over are reaching peak capacity, talked up the wonders of accelerated computing powered by the GPUs that are Nvidia’s bread and butter, and mentioned that even sovereign countries are getting into the AI game. 

“You can see that there’s just so many different directions that generative AI is going, and so we’re actually seeing the momentum of generative AI accelerate,” he said.

A spokesperson for Nvidia declined to comment to Fortune.

The AI tools that are already out there—ChatGPT, AI image generators, and coding assistants—are just the tip of the iceberg. Below the surface are future mainstream applications like extremely targeted ads, recommender systems that predict future wants and needs, and improved search engines, among other things, he added.

But analysts weren’t satisfied with Huang’s answer, and the subject of its customers’ spending was brought up once again. This time, Huang didn’t mince words. 

“The people who are investing in Nvidia infrastructure are getting returns on it right away. It’s the best ROI infrastructure, computing infrastructure, investment you can make today,” he said.

When data center customers—which make up about $26 billion of the company’s $30 billion revenue—already have a trillion dollars tied up in general-purpose computing infrastructure that in Nvidia’s view represents the past, why would they want to build more? Huang asked.

“For every billion dollars with a general-purpose CPU-based infrastructure that you stand up, you probably rent it for less than a billion, and so, because it’s commoditized,” he said.

Data centers that build using Nvidia’s chips for accelerated computing, on the other hand, save money and see increased demand from the growing crop of startups creating the “next frontier” of AI, Huang reasoned.

“The world of general-purpose computing is shifting to accelerated computing. The world of human-engineered software is moving to generative-AI software,” Huang said. “If you were to build infrastructure to modernize your cloud and your data centers—build it with accelerated computing Nvidia. That’s the best way to do it.”

As of Wednesday Nvidia was the second-most valuable company in the world, with its $3.09 trillion market cap second only to Apple‘s. The company reported $30 billion in revenue, two times higher than its revenue in the same quarter last year. Its profit of $16.6 billion was also more than double that of a year ago, and the company said it expected revenue of $32.5 billion for the current quarter, above the $31.9 billion analysts expected.

Despite Huang’s optimistic words, Nvidia’s shares fell just under 6% in after-hours trading Wednesday.