Nvidia dominates the market for the powerful chips that underpin generative artificial intelligence, potentially the most transformational technology since the Industrial Revolution. It briefly surpassed $5tn in market value in October, the first company to do so. Yet author Stephen Witt says this extraordinary success must be tormenting Nvidia\u2019s founder and chief executive, Jensen Huang.<\/p>\n
\u201cIt\u2019s hard to be Jensen day to day. It\u2019s almost nightmarish. He\u2019s constantly paranoid about competition. He\u2019s constantly paranoid about people taking Nvidia down,\u201d says Witt, whose thrilling and detailed account of Nvidia\u2019s rise, The Thinking Machine, was this week named<\/a> Financial Times and Schroders Business Book of the Year.<\/p>\n Since the book was published in April, scrutiny of Nvidia has only intensified, as have potential challenges to its supremacy. In an interview the day after the prize-giving, Witt identifies Google\u2019s rival chips, known as tensor processing units<\/a>, as \u201can almost existential threat\u201d to Nvidia. The latter\u2019s graphics processing units, or GPUs, are the backbone for OpenAI\u2019s ChatGPT, among others. TPUs were used to train Gemini 3, Google\u2019s rival large language model.<\/p>\n Witt says Huang has \u201ctold his colleagues, \u2018Look, you have to understand if you\u2019re working at this company, there\u2019s a team inside Google whose job is to kill us\u2009.\u2009.\u2009.\u2009And they\u2019re smart. They\u2019re great. They\u2019re highly capable and have some of the best engineers. And all they want to do is destroy us.\u2019\u201d<\/p>\n The author is, however, a believer in the vast potential of AI and he knows from his intensive study of Huang and his company, that paranoia is the Nvidia chief\u2019s rocket fuel: \u201cHe\u2019s\u2009.\u2009.\u2009.\u2009driven by negative emotion in a way that I have never seen a CEO before, but he\u2019s able to use this to repurpose his energy into these very productive use cases.\u201d<\/p>\n Witt\u2019s interest in Nvidia was triggered first by the public launch of ChatGPT three years ago, which fed the journalist\u2019s concern about his own future. \u201cI was just like, \u2018Man, I am cooked.\u2019 This thing can write almost as well as I can and very soon is going to eclipse me. What do I do? I guess I\u2019d better just pivot to start writing about AI.\u201d He embarked on a New Yorker article about Nvidia, thinking \u201cit\u2019s just a hardware company. It\u2019s probably had a rotating cast of CEOs. It\u2019s probably had a tortured and maybe somewhat boring corporate history.\u201d Instead, he found \u201cone guy pushing relentlessly for one vision, as an absolute maverick\u2009.\u2009.\u2009.\u2009and then ultimately conquering his whole industry\u201d.<\/p>\n Taiwan-born Huang, now 62, emigrated to the US with his family in the 1970s and showed an early determination to succeed. As a teenager, he trained himself to become an elite table-tennis player, illustrating two key traits, according to Witt: \u201cOne, he can learn so fast. His IQ is so high, his ability to just absorb, process new information and get to a world-class level very, very quickly is just unparalleled. He just has an incredible ability to learn new stuff. And then, two, he has that competitive athlete\u2019s killer instinct. He absolutely hates to lose.\u201d<\/p>\n Nvidia initially existed to serve the fast-growing gaming industry with chips capable of generating better graphics faster, using a process called parallel computing. But in his pursuit of the next big opportunity, Huang took the high-risk approach of catering to what he called \u201czero-billion-dollar market\u201d opportunities, in the hope, Witt says, that \u201csomeday someone will come along and start using your product\u201d. Artificial intelligence based on neural networks \u2014 the foundation of generative AI \u2014 was one such apparently unpromising market when Huang redirected Nvidia\u2019s innovative chips towards it in the early 2010s. It is an approach Witt now expects Nvidia to apply to robotics.<\/p>\n Nobody can argue with Huang\u2019s success, though Nvidia has had to weather fierce attacks from activists, brushes with bankruptcy and several stock-price collapses. Yet some still think good luck played a large part in Huang\u2019s rise. Witt counters: \u201cIf you go out into the ocean with your net and you go stand in some part of the ocean that nobody stands in, and you throw the net into the ocean every day for 10 or 11 years, and then at the end, you catch the biggest fish anyone\u2019s ever seen, did you get lucky? Maybe. But you also put yourself in a position where you could get lucky \u2014 and you were the only one standing there.\u201d<\/p>\n The sheer determination required to sustain that level of risk for that long and survive is vividly painted by Witt in The Thinking Machine<\/a>, which depicts Huang as a compulsive workaholic, obsessed with the best-selling business book The Innovator\u2019s Dilemma, in which management professor Clayton Christensen<\/a> warns how complacent incumbents can fall prey to scrappy smaller competitors. Huang\u2019s proneness to rant publicly at his staff about mistakes is matched with a strange loyalty towards, and from, them. (The fact Nvidia\u2019s stock purchase scheme has made many employees into multimillionaires helps bind them to the group.)<\/p>\n Witt triggered Huang\u2019s famous temper once, at the end of his research quest, when he tried to air AI pioneers\u2019 concerns about the impact of the technology on the future of the species. For more than 20 minutes, Huang attacked Witt after he asked what new jobs AI might create. Nvidia was \u201cnot a manifestation of Star Trek\u201d, the CEO raged. \u201cIt\u2019s just a serious company, and I\u2019m a serious person, just doing serious work.\u201d<\/p>\n In the meantime, Witt is happy to use the tools Huang and Nvidia\u2019s customers have built. In his award acceptance speech, he sent a ripple of nervous excitement around a room full of traditional publishers with a vision of the book as a constantly evolving, AI-fuelled project.\u00a0<\/p>\n In the interview, he wonders whether \u201cthe book could respond to the reader, meet the reader in the middle in some way, understand where the reader\u2019s at in their own knowledge, and then, on the fly, using AI, generate a unique bespoke text that speaks directly to their concerns\u201d. In this world, \u201cthe book moves from being a static printed object to maybe a more dynamic [product], almost like a knowledge database, and I can even use feedback from readers to go do new reporting and answer reader concerns\u201d.<\/p>\n Strangely, his idea echoes a similar concept laid out by Thomas Friedman in 2005, when he was interviewed the day after he won the inaugural FT book award with his paean to globalisation The World Is Flat<\/a>. Friedman suggested that for subsequent editions \u201cwe actually turn the book into an open-source product. Just put it up on the web like Wikipedia and let people add to it\u201d.<\/p>\n There is more than a hint of Huang\u2019s self-motivating fear in Witt\u2019s idea: \u201cI saw so many journalists left behind during the internet era and so many publications just decimated during this shift away from classical print distribution to the internet and I don\u2019t want to live through that myself,\u201d says the author. \u201cI don\u2019t want to be obsolescent.\u201d<\/p>\n