He yells in public for 90 minutes, humiliates his engineers in front of thousands of people and, according to his biographer, does not fire anyone. The same logic governs the way he manages artificial intelligence inside Nvidia: a thousand open projects and not a single one pruned.
The story was told to Stephen Witt while he was writing Jensen Huang’s biography. A microchip left the factory with a defect. Huang called a company-wide meeting in Nvidia’s cafeteria, with thousands of employees present, ordered the chief engineer responsible to stand up and yelled at him for an hour and a half. It was not an isolated incident: Witt himself was subjected to twenty minutes of shouting.
“I am trying to torture you into excellence,” Huang says in his defense.
What happens afterward is more striking. He may demote someone, remove their responsibilities and do it several times to the same person. But he does not fire them. According to his biographer, Huang does not dismiss anyone at Nvidia.
Huang is 63 years old. He founded Nvidia in 1993 in an inexpensive Silicon Valley restaurant with Chris Malachowsky and Curtis Priem. He took the company public in 1999 and still owns approximately 3% of its shares. That stake has given him a fortune of more than $200 billion and placed him among the seven richest people in the world.
Nvidia is worth more than $5 trillion and is the most valuable company ever created. Huang is an electrical engineer educated at a public university, holds a master’s degree from Stanford and is the only major figure in artificial intelligence whose career began in hardware.
A thousand flowers without pruning
When asked where a company seeking to adopt artificial intelligence should begin, Huang rejects the most obvious starting point. Do not start with return on investment, he says. Not because it does not matter, but because it is impossible to calculate at the beginning: no one can enter the performance of a tool they are still learning to use into a spreadsheet.
His method is different.
“Let a thousand flowers bloom.”
At Nvidia, he admits, the number of artificial intelligence projects is out of control, and he means it as a compliment. The company uses everything, from Claude and Codex to Gemini. When someone on his team asks to test a new tool, his first answer is yes. Only afterward does he ask what it is for.
The comparison he uses comes from family life: parents do not demand that their children prove in advance that something will work before allowing them to try it. At work, however, people make that demand constantly, as though they possessed a crystal ball.
He knows that the garden will eventually need to be organized, because a thousand flowers growing simultaneously also create chaos. But he warns against doing it too early: when a company places all its resources behind a single arrow before the right direction is clear, choosing the wrong one means losing everything.
“I have not started pruning yet.”
The other side of that tolerance is a brutal clarity about the center of the business. Huang knows exactly which work sustains his company —chip design, software engineering and systems engineering— and that is where he concentrates its talent.
That is why Nvidia partnered with Synopsys, Cadence, Siemens and Dassault, the suppliers of the tools it uses to design its own chips. Huang wants Nvidia’s technology embedded inside those platforms and is prepared to provide whatever those companies need to make it happen.
“I am going to make sure they have 1,000% of whatever they ask for.”
A tape recorded over
None of this can be understood without Kentucky. The Huang family was Taiwanese and middle class. His father was an electrical engineer, while his mother was a teacher who made him learn ten randomly selected English words every day.
In the late 1960s, the family moved to Bangkok for work. In 1973, amid an uprising against Thailand’s military government, his parents decided to send their children to safety in the United States, where recently immigrated relatives were living. They sold almost everything they owned to pay for the journey and their education.
They enrolled the boys in what they believed was a prestigious educational institution, the Oneida Baptist Institute. In practice, it operated more like a reform school. Jensen was ten years old, had long hair and spoke with a very strong Taiwanese accent.
His brother was sent to harvest tobacco. Jensen was assigned to clean the bathrooms every day. His roommate was a 17-year-old who could not read or write and whose body was covered in tattoos and scars. Huang taught him to read and write; his roommate taught him how to do push-ups.
The only part of that period that Huang says still breaks his heart, however, is something else. The family could not afford international telephone calls, so his parents sent the brothers a tape recorder and a cassette.
Once a month, they sat down and recorded themselves describing what they were doing. They mailed the cassette to their parents, who listened to it, recorded their own messages over it and sent it back. They continued the exchange for two years. Nothing remains of those recordings.
Huang remembers telling his parents that he had joined the swimming team because, after competitions, the students were taken to a good restaurant. It was a brightly lit place that looked like the future to him. It was a McDonald’s.
Almost every year, he sends a video congratulating the graduates of that institution.
“My questions are the most valuable thing I have”
Among the industry’s leading figures, Huang is virtually the only one who has never shared the widespread fear surrounding artificial intelligence. He believes that talking about existential dangers slows progress by relying on hypothetical threats. Witt, who interviewed him for months, believes Huang is sincere and wrong.
Huang is suspicious in another area, however, and that is where one of his most interesting ideas appears. Nvidia built its own artificial intelligence system inside its facilities instead of relying on cloud services.
The reason is not technical.
“My most valuable intellectual property is not my answers. It is my questions,” he explains.
Answers have become a commodity. Questions, by contrast, reveal what someone is thinking about, which problems they are trying to solve and what they consider important. Huang wants those questions to remain inside a small room, behind closed doors.
