Nvidia CEO Jensen Huang discussed artificial intelligence (AI) and more in an interview with Liz Claman on "The Claman Countdown."

Huang, who co-founded the California-based chipmaker in 1993, told Claman that "everybody" is "racing" to adopt artificial intelligence technology. 

Adoption is "incredibly fast" among what the Nvidia CEO described as the "core technology builders, the model builders and the companies that are building consumer-oriented AIs." 

"The frontier models, and all of the startups trying to build a next-generation of frontier models, they’re racing to get to the next level," he added.

Agentic AI is "developing very nicely" and "likely to come on next," according to Huang. He said the next phase after that would be physical AI.

"This is AI that understands the laws of the physical world. It understands things like inertia and gravity and cause and effect and object permanence, and we’ll use that for self-driving cars and robotics systems and things of that nature," he explained.

Huang said such industries "are kind of coming on in waves" and "building on top of each other."

Meanwhile, on the topic of data centers, he said capital expenditures are showing "huge growth" in comparison to last year. 

"Not only is the capex larger, the amount of that capex dedicated to AI is going to be far greater," the Nvidia CEO said during the interview. "We’re seeing a very robust growth for this year compared to last year, and I think we’re teed up for some pretty great build outs of these data centers to produce essentially what I call AI factories, which in the future will transform energy, if you will, into digital intelligence… We’ve got several years, many years of build out to go." 

Huang also discussed DeepSeek, a Chinese startup that released models that it claimed are comparable to leading U.S. ones at a much cheaper cost, and how its technology has affected AI and Nvidia. 

The startup has "opened up reasoning AI" and "opened up a whole new field that consumes an enormous amount" of Nvidia’s computing capability, with many AI developers making use of the new types of models. 

The new generation of AI models has "caused demand to be really high," he said.

During the interview, Huang also weighed in on the sit-down he had with President Donald Trump in late January. The U.S. currently has multiple semiconductor chip controls for exports to China.

Huang told Claman that Nvidia is "committed and very enthusiastic to work with the Trump administration to achieve its goals and want the administration to succeed." 

"But the net of it all is, if you look at where Nvidia’s computers are that are going to China today and you compare it against what we just launched this quarter, this record-breaking quarter that we had with Grace Blackwell, the comparison of the technology, anywhere from 20 to 60 times higher performance is what we currently have in the United States, and so every day that goes by, the export-controlled technology is further and further behind, if you will, in a lot of ways. It’s another way of saying every day that goes by we’re making that technology slower and slower," he continued. "So I think the conversation is good, and we’re happy to share with the administration everything we know, so they can make the best decision." 

His interview with Claman came after Nvidia on Wednesday released its fourth-quarter earnings.

The company generated $39.3 billion in revenue during the three-month period, marking a 78% increase from the same quarter last year. Its net income, meanwhile, came in at nearly $22.1 billion, an 80% year-over-year jump. 

Nvidia’s market capitalization hovered around $2.94 trillion as of Thursday.

TickerSecurityLastChangeChange % NVDANVIDIA CORP.120.15-11.13 -8.48%