32 Years, $5 Trillion — He Did Only One Thing: Bet on a Future Nobody Believed In
Not funding.
Not acquisitions.
Not government subsidies.
Not riding a trend.
Just one man, in a time when everyone said "this is useless," persisting for 18 years.
18 years later, he built his company into the most valuable enterprise on earth: $5.27 trillion.
Let me tell his story first. After that, we'll talk about why 99% of founders bet their time on the wrong things.
In a booth at Denny's, three men ordered coffee and french fries.
April 1993, San Jose.
Jensen Huang sat by the window, across from Chris Malachowsky and Curtis Priem.
Spread across the table were napkins covered in circuit sketches.
They were debating a question nobody could answer —
Beyond gaming, what else could a GPU do?
Jensen Huang said: everything.
Chris laughed. Curtis stayed silent.
Later, they walked out of Denny's carrying three things: $200 in startup capital, a few napkins covered in sketches, and an idea nobody believed in.
That year, Jensen Huang was 30 years old.
Stop here for a moment. I need to make clear what it is he built.
GPU — Graphics Processing Unit.
Sounds like a graphics card. You install it in a computer. You play games.
If you had said that in 1995, nobody would have disagreed.
The entire Silicon Valley saw the GPU as a narrow consumer-electronics accessory — there was a market, but the ceiling was right there, obvious to everyone.
Jensen Huang saw it differently.
He believed the GPU did more than render game graphics — it could do mathematical computation.
Massively parallel computation.
People thought he was talking nonsense. Because saying "parallel computing" in 1995 was like saying "air can fly" in 1880.
But Jensen Huang kept turning the question over: if a GPU can handle thousands of computational tasks simultaneously, why should it only be used for drawing pictures?
He turned that question over for 13 years before making his move.
But first, let me cover 1993 to 1996.
In those three years, he came within a hair's breadth of killing the company entirely.
In 1995, NVIDIA released its first product — the NV1.
Two years of development. Two rounds of funding burned through.
The result?
A disaster.
The NV1 used a proprietary rendering technology that was incompatible with the entire industry.
Sega had placed a large order, but canceled it after seeing the finished product.
Sales approached zero.
Spring 1996. Jensen Huang sat in his office, staring at the layoff list.
He had to cut half his workforce.
From 100 people down to 50.
That night, he sat alone in his office until two in the morning. The next day, he called employees into the conference room, one by one, and told them himself.
He later described that moment of self-doubt in an interview:
"What if I was wrong from the start?"
"What if the GPU really is only good for gaming?"
"What if I've led 100 people to the edge of a cliff, and there's nothing on the other side — how do I bear that responsibility?"
In the end, he made two decisions.
The first: pivot entirely to the Direct3D standard and abandon the proprietary approach.
The second: ask Sequoia Capital for another $2 million.
During the investment committee meeting, the Sequoia partner said a line that has since been quoted countless times:
"We're not investing in a company. We're investing in a person."
April 1997. The RIVA 128 launched.
At Comdex, the line of people waiting to try it stretched from the booth all the way to the exhibition hall entrance.
Four months. One million units sold.
NVIDIA had survived.
But that's not what made Jensen Huang great.
It was what came next.
In 2006, Jensen Huang brought a new idea to the board.
It was called CUDA.
In plain terms — make the GPU no longer just a graphics card, but a general-purpose computing platform.
Scientists could use it to compute protein folding.
Engineers could use it to simulate fluid dynamics.
AI researchers could use it to train neural networks.
Nobody on the board understood what he was talking about.
Investors started calling in: had the money gone to his head?
CUDA's annual R&D investment exceeded $500 million, and the revenue source at the time was —
Zero.
No one needed GPUs for computation. No one at all.
Year one, CUDA's adoption rate was so low it barely registered.
Year two.
Year three.
Year five.
All the way until 2012.
A graduate student named Alex Krizhevsky used two NVIDIA GPUs to train a deep neural network for image recognition.
Its recognition accuracy crushed every traditional algorithm.
AlexNet was born. The deep learning era had begun.
Suddenly, everyone working in AI realized —
The only hardware they needed was called the NVIDIA GPU.
From 2006 to 2012. He had waited six full years.
In 2016, OpenAI received its first NVIDIA DGX-1 supercomputer. Jensen Huang delivered it personally.
He signed the chassis:
"To Elon and the OpenAI team — to the future of computing."
Seven years later, ChatGPT ran on tens of thousands of NVIDIA GPUs.
Another year later, NVIDIA's market cap crossed $2 trillion.
In 2025, it crossed $5 trillion.
The seed planted in 2006 had taken 18 years to grow into the single largest tree in the history of human commerce.
Now stop for a moment.
Reading this far, I want to ask you a question.
How many years have you been building your company?
Have you ever committed to a direction for more than five years — even when there was zero return, when everyone was questioning you?
Not just enduring and waiting. I mean actively investing, steadily doubling down.
If your answer is zero —
Don't rush ahead. Sit with that question first.
I've studied Jensen Huang for a long time.
Here's the plain truth. It comes down to three things. Any founder, in any industry, can replicate them.
First: he didn't chase trends — he waited for the trends to come to him.
I've seen too many founders jump on whatever is hot.
Web3 in 2021. AIGC in 2022. Embodied intelligence in 2023.
A new direction every year, starting from zero every year.
Jensen Huang did the opposite.
He poured investment into CUDA for 18 years. The first six years: completely no return. Year seven: AI arrived.
He wasn't chasing the wind. He was squatting on the road the wind had no choice but to take.
The direction you're on right now — how many years are you willing to squat there?
Second: he didn't do addition. He did multiplication.
CUDA wasn't a new product. It was a layer of infrastructure that made every NVIDIA GPU more valuable.
With CUDA, the GPU went from "gaming accessory" to "computing platform."
Once a platform is adopted, the switching cost is staggeringly high.
Millions of AI engineers worldwide write code that runs on the CUDA architecture.
You want to switch to another GPU vendor? Go ahead. But you'll have to rewrite all your code.
That's the moat — not patents, not pricing, but the ingrained habits of millions of people.
Does your current product have a layer of infrastructure that makes it impossible for customers to walk away?
Third: he turned "technical judgment" — the most intangible capability — into the most concrete competitive barrier.
Jensen Huang once said:
"What I see is not today's market. What I see is that three to five years from now, the GPU will become the core of every computer."
He said something similar in 1999. He said it again in 2006. He said it again in 2015.
Every time, people laughed at him.
But across 30 years, he has never been wrong.
This isn't luck. This is the judgment he accumulated through 30 years of deep cultivation in one direction — GPUs.
That kind of judgment can't be bought with money. Investors can't hand it to you. It can only be earned through time and focus.
So here's the question.
What domain is your judgment in right now?
How many years have you been cultivating it deeply?
Can you see what your customers will need three years from now?
If your industry gets hit by a tsunami like the one AI brought to GPUs —
Can you, like Jensen Huang, be the first to see the waveform on the seismograph?
Those 18 years of CUDA investment — Jensen Huang fired his own bullets to make them happen.
He used 18 years to prove one thing:
Long-term thinking isn't waiting. It's acting earlier than everyone else — and then waiting longer than everyone else.