The Man Who Gave Silicon a Face: A Eulogy for Jensen Huang's Years at NVIDIA

A eulogy for Jensen Huang’s years at NVIDIA

Somewhere there is an old graphics card in a drawer. Its fan has gathered dust. The computer it belonged to is long gone. Whoever bought it probably saved up for it, compared it with a slightly cheaper model and went to bed too late the night they finally got it running. They wanted to play a game. That small, ordinary transaction feels like a good place to begin remembering Jensen Huang’s thirty-three years at NVIDIA. It gives the story back some of its original proportions. Before governments cared who could buy his chips, a kid cared whether one of them could make a game run.

On November 11, 2026, NVIDIA’s board accepted Huang’s resignation as president, chief executive and director, effective immediately. The announcement came the following day. Colette Kress, the company’s longtime chief financial officer, took over as interim CEO. For a man who had spent decades explaining what would happen five or ten years from now, the words effective immediately made for a hard ending. There would be no farewell keynote. The most consequential announcement of his career fit comfortably on a page.

I suspect many of us in semiconductors will remember the week with an untidy mixture of affection, disappointment and embarrassment at how personally we took it. Most of us never met him. We knew the jacket, the voice, the habit of holding up a piece of hardware as though he had been looking forward all week to showing it to us. And perhaps he had. Anyone who has waited months for first silicon knows the pleasure of finally being able to put the thing on a table. He carried that pleasure onto stages long after he had become famous enough to leave the demonstration to somebody else.

There was something generous in that. The semiconductor industry can be a peculiar place to spend a working life. You can give three years to a chip, solve the hardest problem you have ever encountered and find that even the people who love you have only the vaguest idea what you did. Then Jensen would walk out with a board in his hands, turn it toward the audience, and for a little while people outside our trade wanted to know what was on it. That may sound like a small gift beside the scale of NVIDIA’s business. To the people whose work usually disappears under a heat spreader, it meant something.

The company began in 1993 with three founders, Jensen Huang, Chris Malachowsky and Curtis Priem, who saw an opportunity in three-dimensional graphics. Their early ambition was quite specific. They wanted computers to draw things that people would pay to see and play. In retrospect, it is easy to sketch a straight line from those beginnings to artificial intelligence. The people making payroll in the early years had no such line to follow. A customer buying a graphics card had made precisely one promise to NVIDIA. Deliver a good experience today and I will buy this card. Tomorrow was the company’s problem.

Of all the decisions Huang made, CUDA is the one I keep coming back to. He backed the idea that people outside graphics should be able to program the GPU to do other kinds of work. Consider what that commitment looks like from inside a chip company. Someone has to own the compiler. Someone has to write documentation good enough for a graduate student working alone on a Sunday. Someone has to answer a bug report from a researcher whose entire lab might buy two boards. At the next budget review, all those people still need salaries, and the argument for employing them depends partly on discoveries that other people have yet to make. Huang understood why that was worth paying for.

Then, in 2012, AlexNet gave the world a glimpse of what researchers could do with that freedom. A breakthrough in image recognition became part of the history of a company founded to draw graphics. There is a lovely reversal in that. Engineers had spent years teaching machines how to produce a convincing picture. Here were researchers using the same family of machines to work out what a picture contained. The researchers deserve the credit for their discovery. Huang deserves credit for an earlier act of imagination. He had helped put the instrument within their reach and left them room to play something he had never heard.

That is the part of his reputation I hope survives intact. It took judgment to see the opportunity and patience to stay with it. It also took thousands of people whose names never appeared on a keynote slide. Over time, we got into the habit of calling all of that work “Jensen.” It was affectionate shorthand. It was also an arrangement that suited almost everybody. Journalists got a protagonist. Investors got a person whose confidence they could measure. Engineers got an unusually effective advocate. Huang got the authority that came with being the person everybody wanted to hear. This week exposed the cost of that arrangement.

The circumstances of his departure deserve plain words. Huang acknowledged to employees that he had been unfaithful to his wife. On November 8, NVIDIA’s board opened an independent investigation after reports that he had been targeted by a foreign intelligence effort. At the time, the board said it had no reason to believe national security or company secrets had been compromised. Subsequent reporting alleged a separate relationship with an executive who reported to him. The distinction matters. Huang acknowledged infidelity. The reporting about a subordinate raised a workplace question. The intelligence allegations prompted an inquiry whose conclusions were still outstanding when he resigned. His departure settled who would lead NVIDIA. It left those factual questions open.

Of all the people caught up in this, his family has the smallest obligation to explain anything to the rest of us. Investors can ask for disclosures. Employees can ask what their board knew and when. A spouse owes strangers no account of what happened at home, what forgiveness might mean or whether it is possible. There is a particular cruelty in having other people debate the financial consequences of something you are still trying to absorb as a human being. Some part of this story belongs entirely to the people who have to live with it after the headlines move on.

The workplace question belongs to a wider circle. Anyone who has spent time inside a large engineering organization understands how far a chief executive’s authority travels. A passing remark in a meeting can rearrange six months of work. A word of praise can make a career. When a reported relationship involves someone in the CEO’s reporting line, colleagues have a legitimate interest in whether decisions were fair and whether concerns could be raised safely. Employees deserve answers reached through a proper inquiry. They also deserve some consideration from those of us watching from outside. They showed up to build things. This week they had to answer questions about somebody else’s conduct.

What saddens me most is how readily our industry confuses admiration with familiarity. We saw Huang onstage for so many years that we began to think we knew him. We knew something real about his professional judgment. We could examine the chips, use the software, read the papers and see what customers built. His private character was a much larger claim, and we had very little evidence on which to make it. Some of the disorientation people feel now comes from discovering how much of the portrait we filled in ourselves.

I still think there was greatness in his tenure. I think of it in fairly practical terms. If you are running an engineering company and people come to you with an idea that will take years, your easiest answer is to ask for a customer who needs it now. Often that is the responsible answer. Occasionally it kills something wonderful. Huang could recognize some of those occasions. A generation of engineers saw what happened when somebody with the authority to sign the budget said, in effect, keep going. I would hate for the circumstances of his departure to rob a young technical leader of the courage to make that call.

I would hate, too, for the next founder who studies him to come away believing that exceptional judgment in one part of life guarantees sound judgment everywhere else. A board earns its place during the years when the founder is winning and a difficult question feels almost impolite. The time to make it possible for a junior employee to speak freely is years before anybody needs to hear what that employee has to say. Those are quieter responsibilities than choosing an architecture. People build their working lives around them just the same.

Now Colette Kress and the people of NVIDIA have the strange task of going to work after the end of a story that most of the world thought had only one main character. There will still be regressions to debug, boards that refuse to boot and customers waiting for an answer promised on Friday. Somebody will open a waveform, pull a chair over for a colleague and spend the afternoon finding an error nobody else will ever hear about. In that ordinary work lies much of what Huang spent thirty-three years building. It is also where the people who built it alongside him will begin to receive a little more of their due.

And what do you say by way of farewell to a tenure like his? I think you say the things you can stand behind. He helped give scientists an instrument with which to ask better questions. He gave a great many engineers the rare pleasure of seeing the outside world take an interest in their work. He exercised extraordinary judgment in building NVIDIA. At the end, people who had placed their trust in him had painful reasons to reconsider it. Each of those sentences deserves its own space.

I keep coming back to that old graphics card. I like to imagine its owner turning it over once before putting it back in the drawer. There is a little green board, a dusty fan and, somewhere beneath it, a piece of silicon that did exactly what somebody once hoped it would do. For a few years it gave its owner a great deal of joy. Whatever else we learn about Jensen Huang, that small piece of the story was real. And for those of us who have spent our lives trying to make silicon do something worth all the trouble, it is a good place to say goodbye to his years at NVIDIA.

Sources

• NVIDIA, "NVIDIA Announces CEO Transition", press release, November 12, 2026.
• NVIDIA Corporation, Form 8-K, filed November 12, 2026 (event date November 11, 2026).
• NVIDIA, Corporate timeline.
• Business Insider, coverage of Jensen Huang's exit, November 2026.
• The Guardian, live coverage, November 12, 2026.
• Our earlier report: Nvidia's Board Inquiry Into Jensen Huang: The Dated Facts and What They Mean for the Chip Industry.

This is an opinion piece. Statements about the investigation and the reported relationship describe published reporting and company statements as of November 13, 2026; the board's inquiry had not reported its conclusions at the time of writing.