When Rome built its roads, people thought of them as paths. They were not paths. Rome's roads were the nervous system of an empire. Legions marched on them, merchants hauled goods across them, and news traveled through them. Without those roads, the Roman Empire could not have existed.

The same is true of artificial intelligence as Jensen Huang describes it. It is not a technology. It is the road system of a new era.

June 2025, on stage at London Tech Week.

Jensen Huang stood next to the British Prime Minister. He wore his black leather jacket, as always.

He turned to the audience and said:

"There is a new programming language. The name of that language is 'human.'"

Human.

He was not joking.

When computers first entered the world, operating them required specialized knowledge. There was something called machine language. A language made of zeros and ones. Computers understood nothing else.

Over time, programming languages appeared. Fortran, COBOL, C, Python. They were closer to human speech than machine code, but they were still hard to learn. You had to study for years in college. The world was divided between people who could program and people who could not. That boundary looked like a tall wall.

Jensen Huang says that wall is coming down.

Think about it.

If you want a computer to do something for you, what do you have to do now? In the past, you had to write code.

Declare variables, define functions, run loops. If you didn't know these things, you couldn't make a computer do anything.

Now it's different.

You just talk to it.

"Organize this table for me." "Summarize this document." "Schedule my meeting for tomorrow."

The AI understands and executes.

In Jensen Huang's words, the way you ask a computer to do something has become the same as the way you ask a person.

"Just ask politely."

Not many people know C++. Fewer know Python. But everyone knows how to speak. Korean, English, Spanish, it doesn't matter. The language you already use has become a programming language.

But we need to go one step further. The future Jensen Huang envisions is not just "a computer you talk to." What he sees runs deeper.

There is something called an operating system. Windows, macOS, Linux. It is the first program that runs when you turn on a computer. Every other program runs on top of it. Word, Excel, games, all of them. Without an operating system, a computer is an empty shell.

Jensen Huang says artificial intelligence will become the new operating system.

Let me explain what that means.

Right now, we "launch" programs. If you want to use Word, you click the Word icon. If you want to use Excel, you click the Excel icon. Each program exists separately. We move back and forth between them as we work.

In the future, that won't be the case. The very concept of a program will start to blur.

When you say "Write me a report," the AI will pull together whatever capabilities the task requires in that moment. Text editing, table creation, chart drawing. It will combine them and produce the report. You won't need to think, "I should open Word first." The AI will handle that on its own.

When you say "send an email," you don't need to open the email app. When you say "schedule a meeting," you don't need to open the calendar app. Artificial intelligence acts as a bridge between everything.

In June 2024, at Computex in Taiwan, Jensen Huang declared:

"The era of the software factory is over. The era of the AI factory begins."

What is a software factory? It's the way things have worked until now. Programmers gather and write code. They spend months, years building a program. They sell the finished program to people. That's how Microsoft makes Word, how Adobe makes Photoshop.

The AI factory is different. It doesn't sell pre-built programs. Artificial intelligence creates what's needed on the spot. It doesn't search. It generates.

I mentioned the napkin story in the preface. It's an analogy Jensen Huang likes to use.

A napkin is something you use once and throw away. You blow your nose, wipe your hands, jot a note. When it's served its purpose, it goes in the trash. Nobody treasures a napkin.

That's what software will become in the future.

When you ask artificial intelligence something, it creates a small program in that moment. A program built to answer your question. After showing you the answer, the program vanishes. Like a napkin.

Until now, programs had to be "reusable." You built one and a million people had to be able to use it. That was the only way to recoup development costs. Nobody made a program meant to be used once and discarded. It would have been wasteful.

Artificial intelligence changed that equation. For AI, creating a program is like breathing. It costs almost nothing. It's fine to build a program for just one person, just one time. Use it and toss it like a napkin, no loss.

There's an important distinction here.

Jensen Huang emphasizes the difference between a "tool" and "work."

Excel is a tool. It's like a hammer. A hammer doesn't drive nails on its own. A person picks up the hammer and drives the nail. The same way, Excel doesn't analyze numbers on its own. A person has to open Excel and enter formulas.

Artificial intelligence is different. AI is not a tool. AI is the "work" itself.

When you say "analyze this data," AI chooses its own tools. It decides which analytical method to use. It performs the analysis. It organizes the results and presents them. The person just receives the output.

In Jensen Huang's words, artificial intelligence is "a worker that knows how to use tools."

January 2026, CES in Las Vegas.

Jensen Huang took the stage again. The black leather jacket was the same as always. He told the audience:

"Computing has been fundamentally reshaped. You no longer write software. You train software. You no longer run CPUs. You run GPUs."

Not writing but training. That phrase deserves a closer look.

The old way of making a program went like this. A programmer sat down and wrote code line by line. "If A, then do B. Otherwise, do C." Thousands of lines, tens of thousands of lines of such instructions. The programmer had to think through every possible case in advance.

The way we build artificial intelligence today is different. Instead of writing code, you show it data. "This is a photo of a cat, and this is a photo of a dog. Learn the difference." The AI looks at millions of images and discovers the rules on its own.

The programmer doesn't hand over the rules. The AI discovers them itself. Much like a child learning about the world.

Jensen Huang calls this "training." You don't "write" software; you "train" it. So what does it mean for artificial intelligence to become the operating system?

Think about what an operating system does. You press a key and a letter appears on screen. You move the mouse and the cursor follows. You save a file and it's recorded on the hard drive. The operating system handles all of this. We don't notice the process. We just feel that the computer "works."

In the future, artificial intelligence will play a similar role. When we want something, AI understands it and executes. Which program to use, which data to pull, what order to process things in. AI decides all of it. We won't notice the process. We'll just feel that "things get done."

NVIDIA is already moving in this direction.

In November 2025, Jensen Huang said this during an earnings call.

"NVIDIA's architecture and platform are the only framework that runs every AI model in the world. We are in every cloud."

Everywhere. That was how Jensen Huang put it.

NVIDIA chips sit inside Google's data centers. They sit inside Amazon's servers. They sit inside Microsoft's cloud. Most of the computers running artificial intelligence around the world operate on top of NVIDIA's technology. This is not a coincidence.

Think back to the CUDA story covered in Chapter 6.

In 2006, Jensen Huang introduced CUDA. It was software that let GPUs be used for tasks beyond graphics processing. Wall Street laughed. Who would ever use something like that?

Jensen Huang waited ten years. He grew the CUDA ecosystem slowly. Scientists started using CUDA. Researchers cited CUDA in their papers. Universities began teaching it.

Then in 2012, the age of artificial intelligence opened. An AI called AlexNet delivered a dominant performance in an image recognition competition. That AI had been trained on NVIDIA GPUs and CUDA.

Everything changed from that point. If you wanted to do AI research, you needed NVIDIA. There was no other option. The ecosystem built over ten years had become a barrier to entry.

Jensen Huang is running the same playbook again. This time, it is an AI operating system.

NVIDIA is not a company that sells chips alone. NVIDIA sells systems. It sells platforms where chips, software, and tools are bundled into one package.

There is something called NIM. It stands for NVIDIA Inference Microservices. It is a tool that makes it easy to deploy AI models. There is also something called Omniverse. It is a platform for building and simulating virtual worlds. Robots practice in this virtual world before they go out into the real one.

All of this runs on NVIDIA's chips. It is developed with NVIDIA's software. It moves within NVIDIA's ecosystem.

The partnership with Siemens is a good example of where this is heading.

In 2025, NVIDIA and Siemens announced they would build an "industrial AI operating system." It is artificial intelligence that runs factories. From product design to manufacturing to quality control, AI coordinates every step.

Jensen Huang said this.

"Generative AI and accelerated computing have sparked a new industrial revolution. Digital twins are no longer passive simulations. They become active intelligence for the physical world."

A digital twin means creating a replica of the real world inside a computer. You build an exact copy of an entire factory inside a computer. You run experiments in this virtual factory first. If there are no problems, you apply the results to the real factory.

But now these digital twins think for themselves. They find problems, suggest solutions, and carry out optimizations. Artificial intelligence becomes the brain of the factory.

This is what Jensen Huang calls "the integration of OS, programs, and intelligence."

The operating system, programs, and intelligence merge into one. The boundaries disappear.

It used to work like this. There was an operating system, programs sat on top of it, and people operated those programs. The three were separate.

In the future, it works like this. Artificial intelligence plays the role of the operating system. It creates whatever functions are needed on the fly. People give instructions to the AI by speaking. The three melt into one.

There are concerns, of course.

Can we trust programs that AI creates? What if there are errors? Can we use them for anything important?

Those worries are fair. But looking back through history, new technologies have always started out imperfect.

The first automobile was slower than a horse. The first airplane stayed airborne for only a few seconds. The first computer filled an entire room and performed worse than a modern calculator.

Every technology starts that way. It is born imperfect, improves little by little, and at some point changes the world.

Jensen Huang said this at the 2024 World Government Summit.

"For the last 10 to 15 years, everyone has been saying that children should learn computer science. That they should learn programming. Now it's the opposite. Our job is to create computing technology so that nobody has to program. So that the programming language becomes human language. Everyone in the world is now a programmer. This is the miracle of artificial intelligence."

Everyone is a programmer. Consider the weight of that statement.

Until now, technology belonged to the few. People who learned to code, people who majored in engineering, specialists working inside companies. Only they could truly wield the power of computers.

Artificial intelligence tears down that wall.

Anyone who can speak can command a computer. Farmers, fishermen, teachers, students. If you can say what you want, artificial intelligence will make it happen.

Think again of Rome's roads.

Rome's roads bound the empire together. A farmer on the frontier could reach the markets of Rome. A merchant in the provinces could sell goods in the capital. The roads opened opportunity to everyone.

The AI operating system Jensen Huang envisions works the same way. The benefits of technology don't flow only to a handful of experts. They open to everyone. Anyone can bring their ideas to life. With words.

Of course, new questions arise.

If everything runs on artificial intelligence, who controls that intelligence? What happens when the AI makes a wrong judgment? Doesn't the company building the AI end up holding too much power?

Jensen Huang doesn't dodge these questions.

He talks about the concept of "sovereign AI," covered in Chapter 11. Every nation should have its own artificial intelligence. One country's AI must not be dependent on another's.

Whether that alone is enough remains unclear. New technology demands new rules. Writing those rules is not the job of technologists alone. It belongs to all of us.

Jensen Huang left this piece of advice at London Tech Week.

"You will not lose your job to artificial intelligence. But you may lose your job to someone who uses artificial intelligence."

Artificial intelligence is a tool. Or, to follow Jensen Huang's own words, artificial intelligence is "work." But deciding how to use that work still falls to people. Setting the direction for what AI should do is a human responsibility.

So Jensen Huang says: run. Don't walk.

Think back to the napkin from 1993.

When thirty-year-old Jensen Huang scribbled his idea on that napkin, did he know what the world would look like thirty years later? Probably not.

But he started. With an incomplete idea, on a flimsy piece of paper. That became a $5 trillion company. A company that makes the heart of the world's artificial intelligence beat.

We don't yet fully know what the operating system of the future will look like. Whether AI will truly change everything, how fast it will change things, is hard to predict. The change has already begun. And at the center of that change stands a Taiwanese immigrant in a black leather jacket. A man who started with a single napkin. A man now trying to rewrite the very way the world computes.

Jensen Huang.