FADE IN:
“If we had realized how hard it would be, we never would have started. That’s the different thing about starting a company versus keeping one going. You have to be naive enough to start and stubborn enough to continue.” — Jensen Huang
ONE
THE DISHWASHER
EXT. TAIPEI, TAIWAN — DAY — 1972
A bustling Taipei street. YOUNG JENSEN (9) and his older brother walk with their parents to the airport. Their father, a chemical engineer, has arranged for the boys to live with relatives in the United States while the parents stay behind to work. The family cannot afford to emigrate together.
JENSEN'S FATHER
(kneeling)
Study hard. Be respectful. America is where you will have the opportunities I never had. Do not waste them.
Jensen clutches a small suitcase. He does not cry. He nods once, the solemn nod of a child forced to be older than his years.
Jensen Huang was born in Tainan, Taiwan in 1963. He was sent to the United States at age nine with his older brother. Due to a mix-up, they ended up at a boarding school in rural Kentucky instead of with relatives.
INT. ONEIDA BAPTIST INSTITUTE — RURAL KENTUCKY — DAY — 1973
A reform school — not the relative's home his parents intended. YOUNG JENSEN (10) stands in a dormitory with rough-looking older boys. The school is for troubled youth. Jensen is neither troubled nor old enough to be here. But here he is.
A BULLY towers over him.
BULLY
Where are you from, kid?
YOUNG JENSEN
Taiwan.
BULLY
Where's that?
YOUNG JENSEN
Far away.
The bully shoves him. Jensen doesn't fall. Doesn't fight back. Stands there, absorbing the impact, eyes steady. He has already learned something that will define his life: endurance is a form of power.
JENSEN (V.O.) (breaking the fourth wall)
I cleaned toilets. I mopped floors. Every boy at the school had chores. I was ten years old, four thousand miles from my parents, at a reform school in Kentucky. My roommate was a teenager covered in tattoos who smoked and carried a knife. I didn't speak great English. I didn't know anyone. But I learned something there that business school could never teach: nothing that happens to you is fatal unless you decide it is.
INT. OREGON STATE UNIVERSITY — ENGINEERING LAB — DAY — 1983
JENSEN (20) works in an electrical engineering lab. He is focused, meticulous, and clearly the best student in the room. Beside him: LORI MILLS (20), also an engineering student. They are lab partners.
LORI
You're going to burn out the oscilloscope if you keep pushing the frequency that high.
JENSEN
I want to see where it breaks. You only learn the limits by exceeding them.
LORI
That's either engineering wisdom or a recipe for expensive equipment damage.
JENSEN
(smiling)
Both. Want to get dinner?
Jensen Huang married Lori Mills in 1984. She would become an engineer at Qualcomm. They are still married.
INT. DENNY'S RESTAURANT — SAN JOSE, CA — DAY — JANUARY 1993
A corner booth at a Denny's. JENSEN (29), CHRIS MALACHOWSKY (33), and CURTIS PRIEM (33) sit with coffee and legal pads. Jensen taps the table with manic energy. This is the booth where NVIDIA will be born.
Denny's restaurant. East San Jose, California. January 1993. The booth where NVIDIA was founded.
JENSEN
The future of computing is visual. Graphics. Real-time 3D rendering. Every game, every simulation, every scientific visualization — it all needs dedicated hardware. A graphics processing unit.
CHRIS
Intel owns the CPU. Nobody's going to let us put a separate chip on the motherboard without a fight.
JENSEN
They don't have to let us. We just have to be so good that no motherboard is complete without us. The CPU does general computation. Our chip does the pixels. Different jobs. Different silicon.
CURTIS
There are twenty other companies trying this. 3dfx, ATI, S3, Rendition —
JENSEN
And all of them are building for today's games. I want to build for tomorrow's everything.
Jensen looks around the Denny's. This is where he cleaned tables as a teenager. Now he is starting a company in the same restaurant.
JENSEN
(quietly)
I used to be a busboy here. Did I ever tell you that?
CUT TO:
TWO
NEAR DEATH
INT. NVIDIA OFFICES — SUNNYVALE, CA — NIGHT — 1996
Crisis. NVIDIA's first product — the NV1 — has failed in the market. The company has bet on a proprietary graphics architecture (quadratic texture mapping) that the industry has rejected in favor of Microsoft's Direct3D standard. Cash is running out. Layoffs are imminent.
CHRIS
Jensen, we have six months of cash. The NV1 isn't selling. The architecture is wrong. We need to pivot or we're dead.
JENSEN
(staring at the whiteboard)
We scrap everything. The NV1, the quadratic approach — all of it. We redesign from scratch around Microsoft's Direct3D standard. We ship the NV3 in six months.
CURTIS
Six months? That's impossible. A new architecture from scratch?
JENSEN
Then we skip NV2. We go straight from NV1 to NV3. We bet the company on one chip. If it works, we survive. If it doesn't, we die. But we die swinging.
NVIDIA shipped the RIVA 128 (NV3) in August 1997. It sold a million units in four months and saved the company from bankruptcy.
INT. NVIDIA HEADQUARTERS — SANTA CLARA — DAY — 1999
A product launch event. Jensen stands on stage in his signature black leather jacket. Behind him: the GeForce 256. The screen reads: THE WORLD'S FIRST GPU.
JENSEN
This is not a graphics card. This is a Graphics Processing Unit. A GPU. A new category of processor. The CPU handles logic. The GPU handles vision. Every pixel on every screen will be rendered by hardware like this. Welcome to the visual computing era.
NVIDIA coined the term “GPU” in 1999 with the launch of the GeForce 256. It could process 10 million polygons per second.
INT. JENSEN'S OFFICE — NVIDIA — DAY — 2006
Jensen sits with IAN BUCK (30), a Stanford researcher. On the screen: code running on a GPU — not graphics code, but general-purpose computation. Matrix multiplication. Neural network training. Physics simulation.
IAN
We call it CUDA. Compute Unified Device Architecture. It lets developers write general-purpose programs that run on the GPU instead of the CPU. Parallel processing. Thousands of cores working simultaneously.
JENSEN
Show me the speedup.
Ian runs a benchmark. The GPU completes a task in seconds that takes the CPU minutes.
JENSEN
(leaning forward)
This is it. This is what we've been building toward. Not just games. Not just graphics. General-purpose parallel computing. This changes everything.
IAN
The board won't like it. CUDA will cost hundreds of millions to develop. It generates zero revenue. Our gaming customers don't need it. And right now, the number of developers who want to program on a GPU is approximately... zero.
JENSEN
Today it's zero. But someday, it will be every developer on Earth. We invest now. We eat the cost. We build the ecosystem. And when the world catches up to what we've built, we will be the only platform that exists.
JENSEN (V.O.) (breaking the fourth wall)
CUDA was the most expensive bet NVIDIA ever made. For nearly a decade, it generated no revenue. Wall Street hated it. Analysts asked me in every earnings call why we were wasting money on a programming platform for a market that didn't exist. I told them the same thing every time: the market will come. We need to be there when it arrives. They thought I was delusional. I thought I was patient.
CUT TO:
THREE
THE REVOLUTION
INT. UNIVERSITY OF TORONTO — COMPUTER SCIENCE LAB — DAY — 2012
A modest lab. ALEX KRIZHEVSKY, a PhD student, trains a deep neural network called AlexNet on two NVIDIA GTX 580 GPUs. The task: image classification from the ImageNet competition. The network destroys the competition — cutting the error rate nearly in half.
2012. AlexNet, trained on NVIDIA GPUs, wins the ImageNet competition by a massive margin. It triggers the deep learning revolution — and every neural network in the world begins running on NVIDIA hardware.
Jensen watches the results from his office. His face is calm. He has been waiting for this moment for six years.
JENSEN
(on the phone)
Did you see the ImageNet results? The neural network ran on our GPUs. Two of them. Consumer-grade. And it beat everything else by miles. This is the beginning.
INT. NVIDIA GTC CONFERENCE — SAN JOSE — DAY — 2016
Jensen on stage, leather jacket, presenting the Pascal GPU architecture and the DGX-1 — the world's first purpose-built deep learning supercomputer.
JENSEN
I want to personally deliver the first DGX-1. And I know exactly who should receive it.
He walks off stage and, in a now-legendary moment, hand-delivers the DGX-1 to a research group working on AI.
Jensen personally delivered the first DGX-1 to OpenAI, the AI research organization co-founded by Elon Musk and Sam Altman. That system would help train the early GPT models.
INT. NVIDIA HEADQUARTERS — SANTA CLARA — DAY — 2020
The pandemic. Jensen broadcasts from his kitchen — a deliberate, casual choice that becomes iconic. His leather jacket. His kitchen island. His GPU demos rendered from a home studio. The world is locked down, but AI development is accelerating.
JENSEN
(to camera, from his kitchen)
The data center is the new unit of computing. Not the chip. Not the server. The entire data center. And NVIDIA will power every one of them.
Behind the scenes: orders for NVIDIA GPUs are exploding. Every major tech company — Google, Microsoft, Amazon, Meta — is buying thousands of NVIDIA chips to train their AI models. The demand is insatiable.
INT. NVIDIA GTC — SAN JOSE — DAY — MARCH 2023
The AI boom has arrived. Jensen takes the stage at GTC to a standing ovation that lasts three minutes. The audience — developers, researchers, executives — treats him like a rock star. He is, in this moment, the most important person in technology.
JENSEN
We are at the iPhone moment of artificial intelligence. Generative AI is the most transformative technology since the internet. And every bit of it — every ChatGPT response, every Midjourney image, every protein structure predicted by AlphaFold — runs on NVIDIA GPUs. Every. Single. One.
He pauses. The room is silent.
JENSEN
Twenty years ago, we started building CUDA because we believed the world would need parallel computing for problems that didn't yet exist. Those problems have arrived. And we are the only company on Earth with the platform to solve them.
NVIDIA's market capitalization surpassed $1 trillion in May 2023 — then $2 trillion in February 2024 — then $3 trillion in June 2024, briefly making it the most valuable company in the world.
INT. NVIDIA EMPLOYEE KITCHEN — SANTA CLARA — EVENING — 2024
Jensen stands at a stove, cooking for a group of engineers. He does this regularly — the CEO of a three-trillion-dollar company, making stir-fry for his team. The engineers eat and talk. Jensen listens more than he speaks.
YOUNG ENGINEER
Jensen, how did you know? About AI, about CUDA, about all of it? How did you see it coming?
JENSEN
(stirring the wok)
I didn't know. Nobody knows the future. What I knew was the physics. GPUs do parallel computation. Neural networks need parallel computation. It was a matter of when, not if. My job was to make sure we were still alive when “when” arrived.
YOUNG ENGINEER
Weren't you scared? When CUDA wasn't making money? When the board was pushing back?
JENSEN
I was terrified. Every quarter. But I was more terrified of being the company that had the right technology and gave up too early. Patience is the most underrated competitive advantage in business.
CUT TO:
FOUR
THE BOOTH
EXT. DENNY'S — SAN JOSE — DAWN — PRESENT DAY
Early morning. The same Denny's where Jensen was a busboy. The same Denny's where NVIDIA was founded. Jensen pulls into the parking lot in a modest car — not the car of a man worth $100 billion, but old habits.
He walks in. Sits in the same corner booth. A WAITRESS approaches.
WAITRESS
What can I get you?
JENSEN
Coffee. And a booth with a view. I like this one.
He looks around the restaurant. The Formica tables. The fluorescent lights. The same smell of pancakes and burnt coffee from thirty years ago.
He pulls out a napkin and draws on it — a quick sketch of a chip architecture. The next generation. Always the next generation.
LORI
(sliding into the booth across from him)
Still drawing on napkins at Denny's?
JENSEN
It's where I do my best work.
LORI
The most valuable company in the world was founded in this booth.
JENSEN
And it almost died in this booth. Three times. People forget that part. They see the trillion-dollar valuation and think it was inevitable. Nothing is inevitable. Everything is a choice followed by work followed by more work.
He sips his coffee. Looks out the window. The sun is rising over San Jose — the city where a boy from Taiwan scrubbed tables, where three engineers drew chip diagrams on napkins, where a company bet everything on a future that hadn't been invented yet.
JENSEN
(to himself)
I love this booth.
FADE TO BLACK.
NVIDIA's GPUs power over 90% of all AI training in the world. Jensen Huang has been CEO since the company's founding in 1993. He has never stepped down, never taken a sabbatical, and still reviews chip designs personally. His net worth exceeds $100 billion. He still wears the leather jacket. He still cooks for his employees. The Denny's in San Jose is still open.
Suggested Director: BONG JOON-HO — immigrant story, class dynamics, the tension between humble origins and stratospheric success. Visual storytelling that bridges cultures. Suggested Composer: RYUICHI SAKAMOTO — minimalist, elegant, with moments of overwhelming beauty. The sound of patience rewarded.
THE END