LE PERSPECTIVE / THE STORY

For the first time in Nvidia’s 33-year history, Jensen Huang is no longer its CEO. He resigned on Thursday, November 12, and finance chief Colette Kress is stepping in temporarily. The company will find another leader. But replacing the person who built it from the beginning is a different kind of challenge.

Think of Jensen and you probably see the black leather jacket. A chip held between two fingers. A packed room hanging on his next sentence. It’s easy to look at that version of him and forget how unlikely any of it once seemed. Before the AI boom, Nvidia was three engineers with an idea that might not work.

The Bet That Took Decades
Nvidia’s AI success looked sudden to the outside world. Huang spent years backing CUDA and its developers when the payoff was uncertain and investors wanted better returns. The future everyone wanted in 2022 had been under construction for a long time.

Before the jacket, there was a diner

He grew up far from those giant keynote stages. Born in Taiwan, Huang moved to the United States as a child. At 15, he worked at a Denny’s in Portland, washing dishes, clearing tables and serving coffee. Decades later, he still liked to remind people how many cups he could carry. It’s a wonderfully ordinary detail in an extraordinary career.

In 1993, Huang sat down at a San Jose Denny’s with two engineers, Chris Malachowsky and Curtis Priem. They talked about making computer graphics far more realistic. No AI gold rush. No promise of a trillion-dollar company. Just three people who thought computers could become something more.

Young Jensen Huang seated at a modest office desk in an archival photograph

An early office portrait of Jensen Huang. Exact year not independently established. Image: WirtschaftsWoche archive.

And then they nearly lost it. Nvidia’s early graphics strategy ran into a technological dead end, threatening an important deal with Sega. Huang later called those failures humiliating. He admitted the company had made a mistake and asked Sega for help. That decision helped Nvidia survive long enough to try again. Sometimes the brave thing isn’t pushing ahead. It’s admitting you were wrong.

The second chance counted. In 1999, the GeForce 256 helped establish the modern graphics processing unit. Gamers began to know Nvidia’s name. But the really ambitious idea came later: could chips built to draw virtual worlds help scientists understand the real one?

He believed before the world did

In 2006, Nvidia introduced CUDA, giving developers a way to use GPUs for far more than graphics. The next year, Huang pushed CUDA across Nvidia’s chips. It was costly. Investors questioned the spending. For years, the financial results did not justify the ambition. Huang kept the bet alive anyway.

Jensen Huang presenting at Nvidia's GPU Technology Conference in 2009

2009: Huang onstage at Nvidia’s GPU Technology Conference, long before AI became its headline business. Image: NVIDIA.

Imagine being asked, year after year, why you keep investing in something that isn’t paying off. Why not just sell more gaming chips and move on? Instead, Nvidia kept building software, teaching developers and giving researchers new tools. This is the less glamorous part of his story. No packed arenas. Just the stubborn work of making a future possible before there was a market for it.

Jensen Huang speaking about CUDA at GTC 2012

2012: Huang explains CUDA’s expanding reach. Image: SlashGear, GTC 2012 coverage.

Then came a glimpse of what all that work could mean. In 2012, AlexNet used Nvidia GPUs to help transform image recognition. Deep learning was no longer just an intriguing research direction. The tools Nvidia had spent years building were suddenly proving their worth.

One machine. Then an AI era.

There’s a photograph from 2016 that I keep coming back to. Huang is at OpenAI’s San Francisco office, delivering Nvidia’s first DGX-1 AI supercomputer in person. It’s not a giant data center. It’s one machine, one visit, one moment in a much larger story. Looking at it now, you can see how much of the future was still ahead of them.

Jensen Huang alongside the first DGX-1 at OpenAI in 2016

2016: Huang delivers Nvidia’s first DGX-1 to OpenAI. Archival photo shared by Elon Musk; reproduced via Tom’s Hardware.

When ChatGPT arrived in late 2022, the world suddenly wanted extraordinary amounts of computing power. Nvidia seemed to become an AI giant overnight. Of course, nothing about it happened overnight. Behind the headlines were years of processors, CUDA, software, networking and thousands of people whose work had finally met its moment.

Then, in 2023, Huang went back to the Denny’s where it had started. Thirty years after that first conversation, he stood beside a plaque marking Nvidia’s beginnings. There’s something quietly beautiful about that picture. Same restaurant. Same story. Just an unimaginable distance between the young founder who hoped it might work and the man who knew that it had.

Jensen Huang returns to the San Jose Denny's with its commemorative plaque in 2023

2023: Back at Denny’s, with CEO Kelli Valade and a plaque marking Nvidia’s beginnings. Photo: Don Feria/AP Images for Denny’s.

By 2024, Huang was onstage introducing Blackwell, and Nvidia was supplying much of the computing infrastructure behind the global AI race. Those famous keynotes made the future look effortless. His own stories about the early years reminded us how much failure and uncertainty were hidden behind them.

Jensen Huang presenting Nvidia Blackwell hardware in 2024

2024: Huang introduces the Blackwell generation onstage. Photo: Yonhap.

In August 2026, Nvidia reported $96.2 billion in quarterly revenue, including $89 billion from data centers. Those are extraordinary numbers. But the number that stays with me is 33: the years between a conversation over diner coffee and a company whose technology helps power modern AI.

Jensen Huang on stage at Nvidia GTC in March 2026

2026: Huang at GTC, still presenting the next generation of computing

This goodbye is not a simple one

Now comes an ending nobody can honestly present as a happy retirement. Huang has publicly admitted infidelity and apologized to his family. Nvidia’s board has begun an independent investigation following reports of a possible foreign intelligence effort targeting him. The company says it has no reason to believe its secrets or national security were compromised. The investigation remains ongoing, and its outcome should not be presumed.

That matters. People have been hurt, and serious questions remain. A farewell that ignored those facts would not do justice to the truth. But neither would a story that let a painful final chapter erase everything that came before.

And this was never a one-man achievement. Malachowsky and Priem took the first leap with him. Engineers stayed through the hard years. Researchers believed in CUDA before it was fashionable. Developers built whole careers around tools that might never have existed without that long investment.

What made Huang different was not that he always got it right. He didn’t. It was that he could admit a wrong turn, keep the lessons and still commit to a future that was years away. He stayed with the same company from the first risky conversation to the age of AI.

So goodbye, Jensen. This is not the ending anyone would have written for the man in the leather jacket. His mistakes do not undo the technology and the teams he helped build.

One day, someone else will stand on an Nvidia stage and hold up the next chip. The applause will come. And somewhere behind it all will still be three engineers in a Denny’s booth, deciding to give an unlikely idea a chance.

Get stories like this in your inbox every week. Subscribe to The Week in Perspective.

Photo credits (oldest to newest): WirtschaftsWoche archive (early photo; precise date unverified); NVIDIA (2009); SlashGear (2012); Elon Musk archive via Tom’s Hardware (2016); Don Feria/AP Images for Denny’s (2023); Yonhap (2024)