Only Jensen.

November 12, 2026

After thirty-three years, Jensen Huang resigned as CEO of NVIDIA.

He had done wrong. He said so himself, in plain words, to the whole company. Four days later he resigned.

We are not here to argue about that.

We are here because this is not the first time something he built has broken. The first time it was a chip, which is a different kind of wrong, and we know it. But that time somebody chose grace, and everything you know about NVIDIA came after.

Go back to the beginning

San Jose, California, 1993

It started in a booth at a Denny’s.

Three engineers: Jensen Huang, Chris Malachowsky, Curtis Priem. Jensen had just turned thirty.

They had about $40,000 and no name. They had been saving every file as NV, for “next version.”

Three people.
Three points.

Almost everything a graphics card draws is built out of triangles. This was the first one.

1995, and the NV1

Then he bet on the wrong shape.

NVIDIA’s first chip, the NV1, drew the world in curves: quadratic surfaces, smooth and clever. It was a beautiful idea. Everyone else, and then Microsoft, chose triangles.

“After one year of development, we realized our architecture was the wrong strategy.”

Almost nobody built games for curves. And NVIDIA had already promised the same idea to Sega, for its next console.

1996, and a favour

So he went to Sega and told the truth.

As he tells it, he told Sega the chip was the wrong approach.

That they should find another partner. That NVIDIA could not finish the job. Then the hard part: he asked them to pay anyway, in full, because without it the company was finished.

“I was embarrassed to ask.”

Sega had every right to say no.

Sega said yes.

A Sega executive named Shoichiro Irimajiri backed him. In Jensen’s words, that “understanding and generosity gave us six months to live.”

Grace did not do the work. It bought the time.

Six months. Not a rescue. A countdown. He laid off more than half the company. That part is not a fairy tale. Then he bet what was left on one more chip: triangles, this time.

The RIVA 128 sold a million units in four months.

“Confronting our mistake and, with humility, asking for help saved NVIDIA.”

He did not hide that crack.

He put that mistake in a commencement speech. For years he opened staff presentations with the same sentence: “Our company is thirty days from going out of business.” And in July 2026, thirty years on, he thanked Irimajiri in public, on a stage at an arcade in Tokyo.

November 2026

In November 2026, it broke again.

“Leading NVIDIA is the privilege of my life. My responsibility is to serve you and be worthy of your trust. I have fallen short of that responsibility.”

Four days later he resigned.

Somewhere along the way, we forgot he is a person.

It is an easy mistake. The jacket never changed. The keynotes ran like clockwork. After a while you start treating the man like one of his products: benchmarked, rated, expected to never drop a frame.

A machine cannot be ashamed. It cannot be sorry. It cannot lose sleep.

He can. He is human, and every human who ever lived has gotten something badly wrong. This was not a wrong chip. It hurt people, and he said that too. We know.

A mascot cannot let you down. A person can.

He has been a mentor to people he never met: every speech about failing, every retelling of how it started. And a hero since the first time a game ran that had no business running. We never wanted a mascot. We wanted the person, and this is what that costs.