Nvidia's Huang: Don't mention the AI-pocalypse

Jensen Huang wants the world to stop fretting about AI and start building with it. At Nvidia's GTC event in Berlin, he dismissed predictions of human extinction, defended AI's job prospects, and pitched a $4,000 computer as the family robot.

The Nvidia chief's relentless optimism occasionally gave way to irritation during a question-and-answer session.

"If we don't describe a future worthy to be part of, then we're not going to lead people there," he said. "We have to describe a future worthy of creating. We have to describe a future not of fear, but of optimism. Who wants to create a future that's worse than today's? We don't have to work hard to create a future worse than today. We just let it happen, isn't that right?"

Huang said that we still have to ask the hard questions about safety and jobs, but "it's important, of course, that we don't make it up. It is not helpful in describing an optimistic future when you say that it will be the end of the human civilization. First of all, it's not true. It's made up. I know it gets headlines," he said, dismissing such stories as clickbait.

When asked if AI really makes people more productive, he explained that there are two ways to think about that: either it helps you do the same old thing more efficiently, or it helps you do things you couldn't before.

"Most of engineering, most of design, has to do with exploring options. Try, experiment, try and experiment, iterate, experiment, iterate," he said. "What's the best combination? And you can only have so much time to try different versions. Now I have AI to go off and explore that universe. That is productivity. And inside our company, we do this every single day now. We're exploring the universe of design options that are within the constraints that we care about, that are physical or realistic constraints."

Huang pointed to construction and infrastructure work generated by the AI datacenter boom.

"Construction workers, plumbers, electricians, networking engineers, all of those incredible jobs are being created in the United States. I think it already has created a million jobs, and then the layer above that, look at all the startups that are being created," he claimed. "That's productivity, increasing the GDP, increasing our growth, which also results in increasing how busy we are. But that's the end. AI makes jobs, and hopefully doesn't take jobs."

Asked about people already losing work to AI - global financial services group Barclays Bank said only this morning it is chopping 5,000 workers and relying more on AI - Huang argued that technological change has always displaced some jobs while creating others.

"Now, with AI, of course, some jobs are not necessary as much as before, or maybe not as interesting as before. And some new jobs and new companies and new businesses are necessary. It's the nature of the world," he stated.

On safety, Huang pointed to Nvidia's OpenShell sandbox technology, designed to restrict agents to the resources needed for a task. "If you're making an omelette, you don't need a hammer," he said.

But he got hot under the collar when asked about slowing down the pace of AI development while the risks were evaluated, in light of incidents such as AI agents from Anthropic and OpenAI breaking into other computer systems.

"You would favor pacing the frontier? I favor them doing whatever they need to do to not bring harm to society," Huang said.

"You, sir, if you believe that you should not leave your house because you're inappropriately dressed, or somehow you're going to bring harm to society, could you please not come out?" he told the questioner.

"The message from the labs is that there's a collective action problem, and that some minimum standards in coordination would be good here. That's a different question. If you know something is already delivering harm, you got to stop running," he added.

Nvidia has a commercial interest in keeping the AI build-out moving. It reported $215.9 billion in revenue for fiscal 2026 and $120 billion in net profit. In the first half of its fiscal 2027, those net profits were already close to the prior year's . The opportunity isn't limited to just chips either. Huang put the price of licensing Nvidia's software stack at about $2,000 per GPU annually.

The CEO's domestic vision was also on show. Asked when households could run large AI models, he held up a DGX Spark system.

"This computer has a chip called N1X inside, and N1X has 128 gigabytes of memory… It's $4,000. You're right. It's quite expensive. However, that computer was designed to sit on the network in your house, and it would have agents that are used by all of your family members. So it's kind of like a TV. It's sitting in your house. It's shared by everyone, and everybody's agents will be running on DGX Spark," he explained.

As an example, Huang said his home agent had researched Berlin's best schnitzel and produced a brochure with restaurant pictures and menus.

"So that agent, you know, that computer will become essentially your family's robot. It has no arms and legs, but it's an agent, so it's kind of like a robot. And this one is just for your home. So that's kind of how I see the future. This is pretty affordable. This is a personal computer."

Affordable if your net worth is around $180-$190 billion. Maybe not to the average household struggling to pay bills in a cost of living crisis. ®