Sequoia in conversation with Jensen Huang: The computing paradigm is set to undergo a 60-year transformation. You won’t be replaced by AI, but you will be “dimensionally outmatched” by those who know how to “make good use of AI.”

Source: Sequoia Capital

Compiled by: Yuliya, PANews

Editor’s Note: In the past, our data centers were merely storage places for humans to retrieve information; today, computing is moving toward generation. Every word, every image, every video is produced in real time and is highly customized according to the requester’s context. In this sweeping global wave, Sequoia Capital partner Konstantine Buhler held an in-depth conversation with NVIDIA founder and CEO Jensen Huang, discussing the major changes underway in computing technology. Huang believes automation does not lead to unemployment—it instead drives a comprehensive increase in labor demand and upgrades professions themselves. People won’t lose their jobs because of AI, but they may be replaced by those who are good at using AI.

AI Factories and the Generational Leap in Computing Models: From Retrieval to Generation

Konstantine: Thank you so much for being here, Jensen Huang. We’re in the middle of a large-scale AI revolution—its scale and speed could even surpass the Industrial Revolution. You’ve said that what’s happening now is the largest infrastructure build in human history. At the center of this build are AI factories, and the company enabling all of this is NVIDIA. Can you tell us what an AI factory is? Why is it the most worthwhile investment for every business in the next decade?

Jensen Huang: You can understand AI in many ways. The most familiar to the public might be interacting with it through web browsers and chatbots: you give it a Prompt, and it replies with a paragraph. Even if you’ve been using AI for a while, you’ll find that over the past two or three years, its capabilities have evolved in a truly remarkable way.

Two years ago, everyone heard about ChatGPT. At its core, it is computer software that understands what you input. It can sense, comprehend, and transform that information into other content through generation. For example, you can give it a PDF file and ask it to summarize—that’s text-to-text. You can also ask it to generate an image based on a story—that’s text-to-image. Or you can give it a photo and ask it to describe the scene—that’s image-to-text. This ability was called generative AI two years ago.

But beyond understanding and generating, what’s even more valuable is the ability to think. The foundation of generative AI gives it the capacity for internal thinking, step-by-step reasoning, and problem-solving. Not only that—it can now generate control instructions to use tools, whether that’s using digital tools like browsers, spreadsheets, Photoshop, AutoCAD, and so on, or in the future controlling mechanical systems (which is robotics and autonomous driving).

Two years ago, people found ChatGPT interesting—it could write poems and songs—but occasionally it would blurt out nonsense. And today, two years later, we have agentic systems. AI is no longer just about understanding information; it can now reason and do useful work. Because it can do useful work, AI has created real business value. We don’t pay for friends who only talk big—we pay for people who can actually get things done. Now, every day, people are hiring AI by the hour—paying it $20 to $30 per hour. That’s also why it has become the fastest-growing software business in human history.

From the upstream logic of industry, we need to return to first principles. The computer industry we all know today was established about 64 years ago. Back then, IBM launched the System/360, which was also why IBM became the most valuable company in the world at the time.

Over the past 60 years, the essence of computing has been pre-recording and retrieval. You write a story, take photos, record videos, save them as files to a hard drive; when you want to use them, you retrieve them from the hard drive. That’s also why those buildings are called data centers. They mainly store data and don’t do much computation.

But now the situation has changed. In the AI era, every time you provide new background information (context) and a new request, AI performs real-time understanding, reasoning, and generates entirely new results. For example, my speech right now is generated in real time based on the different contexts of everyone in the room—not copied word-for-word from a script. That’s what we call intelligence.

In the future, every pixel, every sound, every video, and even every advertisement and news item will be tailored to you and fully generated, rather than pre-recorded and later retrieved. This means that in the future, we’ll need a large number of generators—large-scale computers we are building—which is what we call AI factories.

The Intelligent Network Encircling the Earth and the Power Generator of the Digital Age

Konstantine: How big will this generator be?

Jensen Huang: At present, we provide information and intelligent generation for about 1 billion people worldwide. But because AI has become agents (Agents), they can do work themselves—an agent can even communicate and collaborate with another agent. Inside NVIDIA, there may be hundreds or thousands of agents talking to each other and solving problems (of course, they all run within safe sandboxes and safety guardrails).

This means that in the future, it won’t be only humans using the internet—there could be hundreds of billions of agents working day and night. Enterprise agents, autonomous vehicles, robots, and even systems inside every building will all be talking to each other. All commands, all thinking, will be generated in real time.

It’s like a thick computational network, wrapping the entire Earth like a cocoon. It sounds exaggerated, but this has happened twice in history:

The first time was 300 years ago. Siemens in Germany built a kind of machine. You turn it on, and it outputs an invisible powerful force—electricity. Today, the power grid (electric grid) already wraps the entire Earth.

The second time was 35 years ago, when the internet was born in the United States. Today, it also envelops global communications.

Now, after energy and communications, we are entering the third network—the intelligent network. NVIDIA’s business that sustains it today is building this new era’s power generator (Dynamo). The power generator 300 years ago took in the physical motion of water flow, wind, or coal (atoms) and output electrons. Ours takes in electrons (electricity) and outputs digital signals. Through different combinations, these digital outputs become languages—language for communication and math, or the language of proteins and human biology; language of physical laws and climate prediction; even the language of a 3D world, robotics, and autonomous driving.

These two kinds of machines, separated by 300 years, are strikingly similar: atoms go in, electrons come out; electrons go in, digits come out. Those digits are what we call Tokens—intelligence. We mass-produce these intelligent Tokens in factories, and that’s the meaning of AI factories.

Konstantine: We’re in a wave where multiple revolutions intersect. From energy transition, to the routers of the global telecom network, to the core of the intelligent revolution—GPUs and AI factories, like the H100 or the latest Vera Rubin architecture. Integrating everything required.

Jensen Huang: Yes. Our computing unit is called a “rack.” A rack has 72 chips. This year, we’re going to build about 8 million such units. A rack weighs 2 tons, costs $4 million, and contains 1.5 million components. It is the most expensive equipment in the world, but we produce them in bulk the way we make smartphones—shipping them to data centers worldwide. It’s huge, and transporting them is definitely labor-intensive.

The Five-Layer Cake Investment Logic of the AI Era

Konstantine: This is a very exciting picture. Whether you’re a large enterprise or an individual, how should we participate in this revolution?

Jensen Huang: If you invest in the AI industry, you can imagine its industrial layout as a five-layer cake. You know, a $50 billion AI factory can generate $300 to $400 billion in smart value—its ROI is incredibly impressive. So what are these five layers?

The first layer is Energy: the bottom-layer power generator. This is the biggest growth opportunity for the energy industry for generations. To support computing, sustainable energy (nuclear, wind, solar, hydrogen, etc.) will receive large-scale investments. As long as energy can be produced, investment will follow. That’s why companies like Siemens, Mitsubishi, and GE Vernova are performing so well now.

The second layer is Chips and Computing Facilities: including chips, computers, network equipment, switches, and silicon photonics technology, and so on.

The third layer is Infrastructure: including land, power, building shells, capital, and the day-to-day operations of data centers. Right now, these resources are in extreme shortage.

The fourth layer is the Models layer: large models built on cloud infrastructure. This is the most investment-dense field in human history driven by the market. People are familiar with OpenAI and Anthropic. But remember: AI can learn not only natural language—it can learn anything structured. We are learning the laws of the physical world. For example, I’m very confident when I sat down just now—not because I had a 47% chance of falling through the chair, but because I believe in physical laws with 100% certainty. AI can also learn the meaning of proteins, genes, and the role of cells. The scale of the physical-world industry reaches $80 trillion—an extremely important frontier, but one that people discuss less.

The fifth layer is the Applications layer: based on the underlying technology, countless startups are reshaping industries such as finance, legal services, accounting, transportation, logistics, and more. Last year, venture capital put $100 billion into this top layer—this was the highest VC investment year in human history.

This future will be enormous. We are just at the beginning—this year, an estimated $1 trillion will be invested into this ecosystem. But my prediction is that in the future, AI will become a massive ecosystem with annual output value as high as $20 trillion. How important is intelligence? Who needs intelligence? How much do you need? Once you figure this out, you’ll know the direction for investment.

AI Is Not Here to Steal Your Job—It’s Here to Help You Upgrade

Konstantine: This isn’t only a multi-trillion-dollar market opportunity and a huge boom in hardware infrastructure and the application layer—it also means creating a large number of real jobs for humanity.

Jensen Huang: Absolutely correct, and we must stress this. Right now, every country and every culture has different attitudes toward AI. But I sincerely want to caution everyone: watch out for the plots from Hollywood sci-fi movies. Don’t always listen to people talking about “the Terminator is coming,” “the technology singularity has arrived,” or “there’s a 20% chance AI will destroy humanity.” That is complete nonsense.

Some even scare people by saying, “We don’t even know how AI works—it’s too mysterious; maybe it will just walk away tomorrow.” That’s even more ridiculous. AI is computers and software. Engineers certainly know how it works—otherwise, how could they keep making it safer and smarter every year?

Today’s AI has significantly reduced hallucinations. The knowledge it generates is accurate and contextually relevant. If it encounters something it doesn’t understand, it will look up information. Before answering you, it even reflects on itself, compares several options, and then tells you the answer. Just like today’s cars are much safer than 100 years ago, the tech community is working incredibly hard to make AI extremely safe.

So focus on what is certain. I’m very sure of one thing: you probably won’t lose your job because of AI, but you will lose it to someone who uses AI.

Since this is a technology that can give humans superpowers, you should get to work using it. Whether it’s telling your loved ones, your children, your company, or your country—embrace AI.

Konstantine: But when it comes to work, people do feel anxious.

Jensen Huang: The moment I hear someone creating panic around jobs, I get furious. This year, we invested $1 trillion into this ecosystem—energy, chips, infrastructure, the models layer, and the applications layer—all creating far more jobs than before.

Someone might ask, what about traditional roles? Here’s a common cognitive mistake people make: they confuse “jobs” with “tasks.”

For example, I’m a CEO. My daily “tasks” mainly involve typing and speaking. Now AI types and speaks far better than I do—it’s at a superhuman level—but as a CEO, I’m actually busier than before.

Let me give you a deeper example. About 12 years ago, a top computer scientist stood up and warned everyone that computer vision analyzing medical images would never get tired—it would never miss any detail. He said it was already at a superhuman level. He predicted that the first profession to be eliminated by AI would be “radiologists,” and advised everyone not to study that major.

He was completely correct in his technical judgment. Today, computer vision is integrated into all radiology systems, and radiologists use AI to assist their work. But what was the result? Demand for radiologists around the world actually increased!

Why? Because the purpose of radiologists is not just to look at images—it’s to diagnose diseases together with clinicians. Due to automation, their efficiency improves, hospitals can handle more patients waiting in line, and radiology becomes more profitable. Hospitals see profits increase and more patients come in, so they end up hiring even more radiologists! People who ignored the warnings and didn’t study radiology ended up missing out.

Similarly, recently some people have said that AI can write code, and with 90% of software programming gone, we no longer need software engineers. But the truth is, we are hiring more software engineers than ever. Because the purpose of software engineers is to solve problems and do innovation—not compete on typing speed. Writing code is just a task; solving problems is the core.

AI won’t eliminate jobs—it will increase the value of your work. If today I were a plumber, I might only follow blueprints to do the job. But with AI tomorrow, I could also be a kitchen designer. If I sell furniture or am a carpenter, in the past you only expected me to nail pieces of wood together—but with AI, I can give you a complete interior design plan, making your home incredibly beautiful. My professional skills are being upgraded!

So I believe that the current narrative that AI will cause human unemployment is completely wrong. It’s just meant to scare others away so the perpetrators can profit from it. Throughout my entire career, computer technology has become increasingly complex. In the past, only 2% of people could master the C++ programming language (maybe more of you in the Silicon Valley venture capital circles understand that). But now, because of AI, as long as you understand human language, you can program. For the first time, we have truly closed the technology divide, and we must bring everyone along into this new era.

LAYOUT REFERENCE (source): total_lines=101, non_empty_lines=51, blank_lines=50