NVIDIA CEO Jensen Huang predicts a 5- to 10-fold expansion in the chip industry and defends Chinese AI models.

Show: Axios Behind the Curtain

Guest: Jensen Huang, Co-founder and CEO of NVIDIA

Duration: 70 minutes

Recorded at: NVIDIA Factory, Fort Worth, Texas

Disclosure: Jensen Huang is CEO and co-founder of NVIDIA, holding approximately 3.5% of the company’s shares and an estimated personal net worth of $181 billion. This content addresses topics such as AI industry expansion, chip demand, and Chinese government regulation, all of which directly relate to his personal financial interests. His position clearly favors promoting AI industry growth and reducing regulatory oversight.

The CEO of the world’s most valuable company sat in his newly built factory in Texas, addressing nearly every major controversy in the AI industry. Huang Renxun spoke plainly: Wall Street’s panic over Chinese AI models is a misunderstanding, AI doomsday theories are “nonsense,” the chip industry will need to expand five to ten times over the next decade, and a bubble is unlikely to emerge within five years. He also publicly stated that Anthropic’s Mythos model should be open to everyone, and that OpenAI and Anthropic will become “the most successful IPOs in human history.” Regarding Trump, he said the president has an excellent memory and can recall chip models like H20, H200, Blackwell, and Rubin; his only concern is that the government might be misled by “science fiction narratives” into overregulation.

The market has misunderstood the impact of DeepSeek, and also misunderstood Kimi. Great models lead to more usage, and more usage means selling more Nvidia computers and building more data centers. It starts with: good models lead to great applications, and great applications lead to growth.

Saying AI will destroy humanity is complete nonsense. Saying AI will eliminate half of America’s jobs is complete nonsense. All facts and evidence point to the opposite.

China has more AI researchers than the rest of the world combined. It is inevitable that China will become extraordinary in this field. Trying to stop China? That’s a foolish idea—and impossible to achieve.

The semiconductor industry needs to expand five to ten times. Today, everything is in short supply—chips, memory, land, electricity, and construction workers. This shortage is fortunately beneficial, as it gives us time to build the necessary infrastructure.

OpenAI and Anthropic will become the most successful IPOs in human history. How can a company worth $1 trillion be only a few years old? Because AI is useful, and when AI is useful, it makes money.

Host Mike Allen opened with the most sensitive topic: FT reported that China is considering tightening export controls on AI models and semiconductors; just after Kimi, a Chinese model, was released, NVIDIA’s stock plummeted, and chip stocks fell 18% in a month. What did Wall Street get wrong?

Jensen Huang’s response was direct and clear. He said the market got it wrong when DeepSeek emerged, and it’s making the same mistake again. “Better models lead to more usage; more usage means selling more NVIDIA computers, building more data centers, and delivering more services—technology will penetrate even more industries.” He repeatedly emphasized this chain of logic: better models lead to better applications; better applications drive growth; growth demands more computing power.

Regarding whether Chinese models like Kimi should be banned, his response was equally direct: "Of course you should use them—it’s smart." He explained that after downloading the model, you can fine-tune it, enhance it, and add safeguards; the model runs within a so-called "harness," which in turn operates inside a "sandbox." The sandbox is secure, with privacy protections, security controls, and access controls. He compared this to an operating system: Linux is open source, reviewed, tested, and hardened by millions of people worldwide, so we can trust it. The same principle applies to open-source AI models.

He also mentioned an often-overlooked point: open models and closed models are not opposites. "The people most likely to upgrade to high-quality models like Anthropic or OpenAI are those who are already using AI." Free models lower the barrier to trying AI; once users get started, they naturally seek better services and are willing to pay for closed-model providers. Therefore, the more open models there are, the greater the opportunities for closed models.

When asked about NVIDIA’s sales in China, Jensen Huang gave an answer many didn’t expect: “Our sales in China today are approximately zero.” He said he has already told investors not to expect any revenue from China, and if the Chinese government and market welcome them back, it would be “a great honor”—until then, he considers sales to be zero.

This was the most tense part of the entire conversation. The host asked: “Are some of your tech peers exaggerating the risks of AI?”

Jensen Huang’s response was blunt: “It’s fine to warn others, and even better to come with solutions—but fabricating facts is absolutely unacceptable.” He then called out several popular AI panic narratives: “Saying AI will end humanity is pure nonsense. Saying AI will eliminate half of America’s jobs is pure nonsense. All the facts and evidence point to the opposite.”

He provided several specific examples to support his argument. The number of radiologists increased by about 20% because AI automation of scan analysis allowed doctors to see more patients—and since there were too many patients needing care, even more radiologists were required. The number of legal assistants increased by about 10%, for the same reason. Manufacturing jobs grew by approximately 50% over the past few years because AI data centers needed to be built and chips needed to be produced.

Intuition and common sense tell you that AI will increase productivity, and increased productivity will create opportunities. Throughout history, technology has made society more efficient, generating more jobs rather than fewer. If that weren’t true, the United States would now have only about 100,000 jobs.

He also turned his criticism toward AI leaders. "Doomsayers spend too much time theorizing sci-fi scenarios, which may make them seem smart." When pressed whether he was referring to certain AI company CEOs, he did not deny it, but said: "If the goal was to make the world aware of this technology’s incredible capabilities, that goal has already been achieved. We should focus our time on ensuring the technology is safe—that’s the responsibility of technology leaders."

The host noted that Asia’s attitude toward AI is vastly different from that of the U.S., where Huang Renxun is mobbed by fans seeking autographs. His explanation was thought-provoking: “Perhaps it’s because doomsayers spend too much time crafting sci-fi endings.” In Asia, people embrace AI as a tool and an opportunity, rather than fearing it as a threat.

The host posed a sharp question: Every industrial revolution has experienced a bubble—where is the bubble risk in this generation?

Huang’s response was cautious but clear: “Bubbles will eventually come, but not today. We are at the very beginning of construction.” He provided a timeframe: unlikely within five years, and uncertain between five and ten years. The reason lies in comprehensive constraints on the supply side. “The industry can now build faster, but we lack chips, memory, land, power, and even construction workers. We are constrained in every direction and on every front.”

This constraint is actually a good thing, he said. "This constraint has held the system back, giving us ample time to build out the infrastructure." Demand is strong, but the ability to convert that demand into productive supercomputing has been delayed due to all these physical limitations. This has pushed back the point at which supply would exceed demand.

More importantly, his assessment of the overall scale of the semiconductor industry: "I believe the semiconductor industry needs to expand five to ten times." Over what time frame? "Over the next decade." This means today’s chip industry is still far too small and insufficient to meet the demands of building AI infrastructure.

He explained why this semiconductor cycle is different from previous ones. "This time is different because it’s not demand-driven, not seasonal, and not consumer-driven. It’s driven by industrial infrastructure. Just as the world needs energy, the internet, highways, and railways, it now needs a layer of intelligent infrastructure for AI, built on top of all existing infrastructure. This layer requires chips."

Regarding the risk of customers borrowing money to purchase NVIDIA products, he said he isn’t overly concerned. “These are exceptional companies that generate massive amounts of cash.” He specifically highlighted the launch of the AI profit flywheel: AI is useful, so AI can now generate profits. Coding agents are extremely profitable, performing valuable work in high-salary roles, and many companies are willing to pay hundreds of millions of dollars annually to enhance their coding capabilities. “This flywheel has already started.”

This is the most technically in-depth part of the entire conversation. The host asked: What convinced you that tokens will become increasingly profitable?

Huang Renxun’s explanation is vivid: “A token is an embedding—it embeds knowledge and intelligence. This number isn’t static, unlike pi. The intelligence encoded in this number grows smarter over time.” As intelligence becomes smarter, it becomes more useful; the more useful it is, the more valuable it becomes; and the more valuable it is, the more people are willing to pay for it.

He compared the AI industry to the past software industry. The traditional software industry was "light capital," which is why software companies had high gross margins. But in the AI era, the software industry will become more "capital intensive," because producing modern software requires machines like the supercomputer in front of him to generate intelligence. Every industry will become more capital-intensive, but the returns will be incredible intelligence, productivity, and growth.

We are laying the foundation and building infrastructure—the largest industrial infrastructure project in human history.

In response to questions about whether AI has already peaked, he said, "It’s impossible to have peaked, because AI’s penetration into society and industries has only just begun." He also noted that over the past six months, $300 billion has been invested in venture capital and startups in the United States alone, creating new jobs and new companies.

Huang’s evaluation of Trump was unexpectedly specific. “He’s smart; he remembers everything. My goodness, he really understands numbers.” He noted that Trump is the only president who can recall NVIDIA chip models like H20, H200, and Blackwell, and even knows that the next-generation product is called Rubin.

He recounted his first meeting with Trump: "He said he wanted to restore America's manufacturing capabilities, re-industrialize the country, establish secure and resilient supply chains, and bring semiconductor manufacturing back to the United States." He said that the Fort Worth factory where they sat for the interview was a direct result of that conversation.

When asked what mistakes the government should avoid, Huang made a rare display of anxiety. "I'm worried about overregulation and overcorrection," he said, noting that some companies want the government to help craft regulations favorable to them. "I believe we should have open competition." He dismissed as "nonsense" the notion that the AI race is like a 100-meter sprint: "Does the first one to cross the finish line win forever? That’s nonsense. The ultimate victory depends on whether society adopts the technology, not who invented it. We didn’t invent electricity or manufacturing, but America adopted them faster and with more enthusiasm—and that’s why America became what it is today."

The host pressed further: “What if Trump calls and asks for NVIDIA’s equity—what would you say?” Huang’s response was clever: “It’s unnecessary. The United States already owns equity in NVIDIA. Last year, we paid $10 billion in taxes, and this year we’ll pay even more. We create vast numbers of jobs and tax revenue. And don’t forget, most Americans today have investments in the stock market—when the market rises, everyone benefits.”

Regarding whether Anthropic’s Mythos model should be open to everyone (currently only available to specific institutions), Jensen Huang’s stance is very clear: “Of course it should be open to everyone.” He said it is Anthropic’s responsibility to ensure the technology is secure, just like any software—when vulnerabilities are found, they should be fixed as quickly as possible. He referenced the Mythos jailbreak incident, saying, “Everything’s fine—we’re still chatting here.”

Regarding the issue of open-model companies distilling closed models and reselling them, his stance is balanced: "It depends on the terms of service. If the service provider is unhappy, they should contact that company. We have many traditional legal remedies available." But he emphasized that learning from other sources is a fundamental characteristic of intelligence. "AI must learn from something. Early AI scraped all existing knowledge from the internet. Now, AI-generated content exceeds human-generated content, and in a few years, 99% of content on the internet will be AI-generated. You’re already constantly distilling the intelligence of other AI systems."

The host asks: When will the ChatGPT moment for bots arrive?

Jensen Huang gave an unexpected answer: "The ChatGPT moment for robots has arrived." He explained that the ChatGPT moment isn’t when AI becomes useful—it was merely seen as "interesting and surprising" when ChatGPT launched in 2022. True usefulness took another four years. Robots are now at that "interesting and surprising" stage. You can tell a robot, "Put the apple in the drawer," and it will reason through the sequence of actions, including opening the drawer first before placing the apple inside. "When you first see a physical robot actually do this, it opens up your imagination for the future of robotics."

Regarding when it would become useful, he provided a time frame of "three to four years" and said, "I wouldn't be surprised."

Regarding the Age of Agents, he painted a striking picture: today, a billion people use computers, but most of the time, those computers sit idle. In the future, everyone will be assisted by numerous agents that use computers around the clock. "We’ll have tens of billions, even trillions of agents running nonstop—intelligent agents, less intelligent agents, specialized agents, super agents—all operating 24/7. So the number of computers we need will grow dramatically."

The conversation returned to Jensen Huang himself. The host mentioned that he founded NVIDIA at age 30, and now the company has a market capitalization of $5 trillion with only 50,000 employees—he said that in ten years, it might still only have 75,000, “as small as possible.”

Jensen Huang defines the role of CEO as "strategy." "Strategy is achieving your future vision as efficiently as possible with limited resources." He says he began this job at age 30, making him likely the longest-serving CEO in tech history. "This is my craft, this is my skill."

Regarding the theory that “suffering and hardship are the keys to greatness,” he clarified that it has nothing to do with any specific painful event, but rather refers to a continuous state. “No great athlete becomes great by accident. It’s the countless hours of practice when no one is watching, the countless failures, the countless moments of suffering and hardship.” He said this kind of hardship refines skill, shapes character, and builds confidence and resilience. “Just as athletes say, when pressure and tension are at their peak, time seems to slow down. I feel the same way. It comes from practice—from doing the same thing over and over again.”

When asked what he would say to his nine-year-old self upon arriving in the U.S., his response was an enthusiastic tribute to America: “America is the greatest country in the world, period. Even though we face challenges and divisions, it is precisely those challenges and divisions that make us great—through open dialogue and a dynamic system that enables freedom: freedom of speech, freedom to innovate, freedom to start businesses, and freedom to collaborate.” He concluded, “This nation was built by immigrants, and it will continue to need incredible immigrants in the future.”

Regarding his habit of not wearing a watch, his explanation was very Jensen: "Because right now is the most important time. I refuse to let my schedule manage my life, or let a watch manage my life. If I’m late, someone will tell me. Until then, I’m 100% here."

Organized & Compiled by Deep潮 TechFlow