Jensen Huang says NVIDIA will sell a few trillion dollars of computing over the next several years

The most expensive pep talk in tech right now is happening on a sidewalk in Taiwan. Jensen Huang told the Lex Fridman Podcast that NVIDIA is preparing to “build and sell a few trillion dollars of computing” over the next several years, and that Taiwan’s tech ecosystem will need to pour “hundreds of billions of dollars” into making it real.

That is not a normal CEO sound bite. That is a capital plan dressed as gratitude, a hype cycle with a bill of materials. Huang wasn’t merely thanking suppliers after a keynote. He was telling an island full of chip-adjacent companies: congratulations, you’re all drafted.

If we’re going to go go build and sell a few trillion dollars of computing over the next several years, this entire island has to be part of it. This entire ecosystem have to be part of it. Every single one of these companies are going to have to invest hundreds of billions of dollars over the next several years.

The island isn’t scenery

This is the part of the AI boom that tends to get flattened into stock charts and keynote leather jackets. NVIDIA may be the face of the GPU gold rush, but the physical reality runs through factories, packaging, memory, boards, power, cooling, logistics, and the brutal choreography of suppliers that make modern chips possible. Taiwan is not a backdrop in Huang’s telling. It’s the machine room.

The claim is big enough to sound absurd, but not big enough to dismiss. NVIDIA has become the tollbooth for AI ambition, and every hyperscaler, startup, research lab, and sovereign AI project wants more compute than the current world can comfortably provide. “A few trillion dollars” is still CEO arithmetic, the kind of number that arrives wearing a varsity jacket and carrying a hockey stick chart. But the logic underneath is blunt: if AI demand keeps rising, hardware spend won’t be a side budget. It’ll be infrastructure.

I was super nervous. But I’m always nervous.

That admission is the useful tell. Huang is not playing the serene prophet here. He sounds like a man who knows the pitch has to land because the next phase of NVIDIA’s dominance depends on other companies spending terrifying amounts of money before the future fully arrives. The keynote wasn’t just theater. It was procurement vibes.

The gamer line is doing real work

Huang also made sure to touch the old altar: gaming. Before NVIDIA became the company that made Wall Street say “accelerated computing” with a straight face, it was the graphics-card company for people who cared deeply about frame rates, heat sinks, and whether their rig sounded like a leaf blower.

Yes, I love the gamers. That’s where we all began. You know, don’t forget your roots.

That line could be pure fan service, but it also helps Huang square the company’s identity problem. NVIDIA is now an AI infrastructure giant, the kind of company that makes governments and cloud providers sweat. Yet it still wants the emotional glow of the gamer kid unboxing a graphics card. The trillions are easier to swallow when they come from the guy who remembers where the fans came from.

If Huang is even half right, the next several years of AI won’t be decided only by model cleverness or chatbot polish. They’ll be decided by whether enough companies in the NVIDIA orbit can finance and build the physical compute stack fast enough. Your cheap magic internet robot has a supply chain, and Huang just put a trillion-dollar price tag on it.