Jensen Huang has left Nvidia, the company he cofounded 33 years ago, closing out a tenure in which he became the public face of the GPU maker and of the broader AI buildout. His exit follows two chip smuggling scandals the company seemingly enabled and allegations of three extramarital affairs, one of them with a Chinese spy while he was mingling with President Donald Trump. The departure raises questions about the US AI industry and the stock market, because Nvidia was the first $5 trillion company and alone represents around 6 percent of the S&P 500 index. Huang, 63, had made no plan for succession.
That gap is difficult to fill because Huang had no deputies, with roughly 60 employees reporting directly to the CEO. Tom Mainelli, an IDC analyst and former computing journalist, said Huang built the company around himself and that everything there flowed through him. Jon Peddie, a computer industry analyst who has tracked Nvidia since its inception, said Huang and Nvidia were one and the same. Pat Moorhead of Moor Insights & Strategy called Huang both the visionary and the tactical leader, the closest thing tech has had to Steve Jobs.
Huang's style drew comparisons to Jobs, from keynote product reveals to a single recognizable piece of clothing, in his case a black jacket that was sometimes a Nvidia-branded softshell rather than leather. He wore it even in hot temperatures and was seen sweating heavily rather than removing it, and he sometimes held keynotes without much news, restating past accomplishments. Unlike Jobs, he lacked the same gravitas, but he had technology to show and demonstrated it onstage, from gaming graphics, physics and lighting to image recognition, AI weather simulations, digital copies of real-world factories and robots trained virtually inside them.
Nvidia's rise came through a series of large bets. In 1996, three years after Huang cofounded the company at a Denny's diner, Nvidia nearly went out of business over a failed graphics chip for the Sega Dreamcast, yet Huang personally persuaded Sega to invest $5 million, The Wall Street Journal reported in 2024. He then bet the company on the Riva 128 PC graphics accelerator, laying off over half of Nvidia's roughly 100 staffers, skipping prototyping and rushing it to production with one month of payroll left, The New Yorker wrote in 2023; it sold a million units in four months and led to both the GeForce GPU and a long relationship with TSMC.
Huang later staked the company's entire gross profit, he told podcaster Lex Fridman, on CUDA, a feature letting a GPU handle computing beyond graphics that added 50 percent to the cost of every GeForce card. Peddie said there was no market demand for it and that Huang had the vision and courage to pursue it anyway. Dave Altavilla of HotTech said Nvidia was investing in a computing platform far beyond graphics with no guaranteed near-term return, and Mainelli said CUDA is why Nvidia was ready when AI arrived.
After the Great Recession of 2007-2009 and design defects in Nvidia laptop graphics chips cost the company 80 percent of its market value, Huang kept investing in CUDA, Altavilla said, handing out high-end Nvidia GPUs to researchers at industry conferences and seeding universities such as Stanford with hardware. By 2012, University of Toronto researchers used Nvidia GPUs to train the prize-winning image-recognition neural network AlexNet, widely credited with sparking the AI revolution. Not every vision landed: Huang said that demo led to his bets on chips for self-driving cars, though CUDA had been conceived as more affordable supercomputing for researchers, and a plan to put Nvidia chips in phones did not pan out.
More company and startup news from TechManNews.
