Nvidia’s Jensen Huang: The Ultimate Entrepreneur

In our commentaries over the past several years, we have written about five of the greatest American entrepreneurs of the recent past: Bill Gates (Microsoft), Steve Jobs (Apple), Mark Zuckerberg (Meta Platforms, formerly Facebook), Elon Musk (Tesla among other companies), and Jeff Bezos (Amazon). These men have surely done more to influence our culture and our country than many of our political leaders over the past three decades. This investment commentary tells the story of another amazing modern entrepreneur: Jensen Huang, president, cofounder, and chief executive officer of Nvidia. When all is said and done, Huang may change the world even more than the PC and iPhone have.

In 1992, Jensen Huang met frequently at a Denny’s restaurant in East San Jose, California with Curtis Priem and Chris Malachowsky. The result of these tentative meetings was the startup company, Nvidia. Thirty–two years later, Nvidia vies with Apple as the company with the largest market capitalization in the world. As this commentary goes to press, the market capitalization of Nvidia is $3.34 trillion, compared with Apple’s at $3.58 trillion. In June 2024, Nvidia claimed the prize as the company with the largest market cap and currently it is a race between three companies for the lead: Apple, Nvidia, and Microsoft. The purpose of this investment commentary is to explain how a company that produced graphics cards for video games in the 1990s came to be one of the three most valuable companies in the world.

In two words, the reason for this is a Taiwanese–American business leader: Jensen Huang. As elucidated below, he masters three skills: technology, business strategy, and operating a business. Very few entrepreneurs possess all three of these skill sets. Most of the information in this commentary comes from an excellent book by Tae Kim: The Nvidia Way – Jensen Huang and the Making of a Tech Giant. Tae Kim writes for Barron’s and Bloomberg; previously he was an equity analyst in the technology sector. The book not only describes the trajectory of Nvidia over the past 32 years but is also an intimate portrait of Jensen Huang. It is highly technical as well, and this commentary is intended for those without the time or interest to plow through the book but are nonetheless interested in Jensen Huang and Nvidia’s remarkable story.

Jensen Huang was born in Taiwan in 1963 to Taiwanese parents. When he was four years old, his father visited New York City and fell in love with America. Starting then, his parents’ goal was to raise their two sons in America — the land of opportunity. His parents were not wealthy and moved frequently because of his father’s work requirements. Huang’s mother taught him and his brother English by selecting ten words a day from the dictionary and having them memorize their definitions. Huang’s parents, while waiting to immigrate to the U.S., sent both brothers to Tacoma, Washington to live with their uncle and aunt, who had recently moved to the U.S. Huang’s parents wanted him and his brother to get a good education and sought a boarding school for them. They landed on one called Oneida Baptist Institute located in eastern Kentucky, as the school took international students. In order to finance it, they had to sell almost all of their possessions. However, the boarding school, which had around 300 students, was not at all what they thought it was. It was not a prep school, but rather a reform school for troubled youth. The students were held to a very strict routine, and they were required to work part of each day. Due to his small stature and his ethnicity, Huang was bullied — even by his roommate, who was eight years older and covered with tattoos and scars from stab wounds. Eventually, Huang befriended his roommate and helped teach him to read, while his roommate introduced him to weightlifting, which helped him stand up to the bullies. Huang would later say that his experiences at school developed in him his streetfighter mentality. He is quoted as saying, “I will never start a fight, but I will never walk away from one.” His days at OBI remind the reader of Elon Musk’s early years in a school near Pretoria, South Africa as recounted by Walter Isaacson in his book on Elon Musk, and described in one of our earlier commentaries.

After a few years, Huang’s parents immigrated to Beaverton, Oregon, located near Portland. They withdrew him and his brother from Oneida Baptist Institute and sent them to public school. In his early teens, Huang developed a passion for ping pong and became one of the best young players in the Pacific Northwest. When he was 15, he entered the U.S. Open Junior Doubles tournament and did well. As a teenager, he worked at Denny’s to earn spending money, made friends at school in the math, computer, and science clubs, and spent all his free time programming BASIC on Apple II and playing video games. Graduating from high school at 16, he enrolled in Oregon State University, receiving his B.S. in engineering in 1984. Shortly thereafter, he married his high school sweetheart, Lori Mills. Getting job offers from LSI Logic and Advanced Micro Devices, he chose to go with AMD, where he helped design microchips while taking night classes at Stanford to get his Master’s degree. After a few years, he moved to LSI Logic, where he was assigned to projects with the start–up company, Sun Microsystems. There he worked with two engineers, Curtis Priem and Chris Malachowsky. Huang had overcome a lot to get this far, and he was just getting started. Later, when asked for advice on how to achieve success, his answer has always been: “I wish upon you ample doses of pain and suffering.”

At Sun Microsystems, Priem and Malachowsky were assigned to work on a SPARC workstation project to create a graphics accelerator card. This would entail moving graphics processing out of the central processing unit (CPU) and onto a separate chip. They succeeded in creating a chip, called GX, for the workstation. At some point later, Sun management wanted to disband their team and perhaps let them go. Instead, they were assigned to build a graphics chip for Samsung, and they reached out to Huang at LSI Logic for help in writing a contract. At this point, the realization came to all three of them: Why not build a great graphics chip ourselves? After many meetings at Denny’s in East San Jose (which Huang knew well, having worked as a dishwasher at Denny’s as a teenager), the three of them decided to take the plunge and resign from their companies to start the new company. Before doing so, Huang had to become convinced that there was an addressable market of $50 million for the chips they intended to produce.

The team started working out of Priem’s townhouse on a chip they called NV1 (NV for new vision). About a dozen others joined them and started working without salaries, as funding had not yet been secured. It took longer to come up with a name, and they finally settled on Nvidia — a combination of invidia, a Latin word for envy, as they hoped that the company would be envied in the future, and NV, because that was what they were calling the new chip they were developing. Nvidia was incorporated with $200 from each of the three founders in April 1993.

Several months later, Nvidia received $2 million in funding from two VC firms. The iconic Don Valentine of Sequoia Capital invested $1 million, and Sutter Hill Ventures did the same. LSI’s CEO, Wilfred Corrigan, played a key role in Valentine’s decision to invest in Nvidia. Corrigan was so impressed with Huang that when he resigned, Corrigan not only asked if he could invest personally in the new company, but he also called Valentine to inform him about Huang. The lesson Huang took away from this is that character and reputation matter most in life.

As Huang recounts, it was a near miracle that Nvidia survived the first decade. Nvidia was initially in the business of producing graphics cards, also known as video cards, which are a computer component that display images, videos, and animations on a monitor. These hardware expansion cards work with the CPU to process and output visual data. In its first several years, Nvidia first produced a graphics card called NV1 and followed it with an upgrade named NV2. Though they were loaded with new topnotch technology, they were not what the game manufacturers were looking for or needed. The firm hadn’t listened to its client base. By the end of 1996, Nvidia was running out of cash and was within months of bankruptcy. Huang had to cut the staff from 100 to 40 to survive. This is why Huang for many years afterwards routinely began every all-hands meeting of employees with the statement that “we are within 30 days of going out of business.”

Nvidia was, however, saved by the release of its next chip — the RIVA 128 (short for Real-time Interactive Video and Animation accelerator). The RIVA 128 was the first Nvidia product to integrate 3D acceleration plus traditional 2D and video. It was a huge success. It was named “Product of the Year” in 1997 by PC Computing and enabled Nvidia to chalk up its first quarterly profit ever ($1.4 million). Several years later, Nvidia went public at an IPO price of $13 a share. It was still a small company; annual sales for the year were only $158 million.

During the internet bubble at the end of the 1990s and early 2000, Nvidia prospered, and its stock price by the end of 2001 had advanced twentyfold. Part of the reason for the stock price spike was that Nvidia had announced the production of a chip which it called a GPU (graphics processing unit). Though Nvidia’s GPU was originally designed to accelerate the rendering of 3D graphics, developers began to tap the power of the GPU to dramatically accelerate additional workloads in high-performance computing and deep learning. Once again, Nvidia introduced a chip, the NV30, which was not well-received by its clients. As Huang later commented, “NV30 was an architectural disaster … a tragedy. The software team, the architecture team, and the chip design team hardly communicated with each other.” Fortunately, Nvidia had accumulated cash reserves, and thus it survived this debacle, but only barely. Nvidia’s stock price dropped 87% in 2002. But Nvidia did not fold during this tumultuous decade — unlike most of the companies in this sector which failed.

The next major event in the story of Nvidia was the development and introduction of the CUDA platform (an abbreviation for Compute Unified Device Architecture) in 2007. CUDA essentially is software which allows developers and software engineers to use a CUDA-enabled GPU for general-purpose processing and computing purposes. Thus, it was software rather than hardware that transformed the company. It became a foundational technology. At the same time, Nvidia introduced the G80, its first GPU chip with a computing function. In what is called parallel processing, CUDA enables hundreds or even thousands of computing cores on a GPU to perform simultaneous calculations, vastly outperforming CPUs on tasks that can be parallelized. It took four years and almost $500 million in R&D expense to create the G80 chip. But the investor community was not convinced at the time that there was a market for the chip. During the 2007-2008 financial panic, Nvidia’s stock again fell more than 80%. But Huang had the ability, like Jeff Bezos and Steve Jobs, to invest strategically for the long term, and it paid off. Nvidia’s GPUs only run in conjunction with CUDA. Today there are more than 5 million CUDA developers, 600 AI models, 300 software libraries, and 3,700 CUDA GPU‑accelerated applications. Moreover, there are about 500 million CUDA-capable Nvidia GPUs in the market.

From the start, Huang believed in giving direct feedback, often very blunt, to his employees in order to reinforce his principles and ensure that everyone had a clear idea of what was expected of them. The following are some of the key principles upon which he has built Nvidia:

• Moving at maximum speed — what he calls the “Speed of Light” — while demanding
  topnotch quality (top speed because the competition in this sector is fierce)
• Giving large amounts of independence for each employee but demanding the highest
  possible standards
• Requiring superhuman levels of effort and mental resilience
• Focusing relentlessly on the mission of the company
• Creating a flat organizational structure to avoid empire-building and internal politics
  (Huang has 50-60 direct reports)
• Demanding public accountability for problems and issues, leading, where he deems
  necessary, to public embarrassment and humiliation
• Operating in teams and informing your teammates if falling behind so they can help
• Demanding intellectual honesty, acknowledging mistakes, and learning from them

From the beginning, Huang created an intense and extreme work culture at Nvidia. Understanding that it was in an incredibly competitive industry with continual new technological advances in chips, he really did believe that Nvidia could well fail unless his associates worked at 110% capacity. Observers often tell of driving by Nvidia’s parking lot at 7 pm or on weekends and the lot being completely full. And Huang drives himself the same way that he does his employees. At one of the sales meetings, the first words out of his mouth were: “I look in the mirror every morning and say, ‘You suck.’” His philosophy is: “Do your job. Don’t be too proud of the past. Focus on the future.”

Huang does not believe in praise; he does not say “great job” or “good work.” No “attaboys.” He believes that praise is a distraction and leads to the deadliest sin of all, which is resting on past accomplishments as if they will protect you from future threats. When things go really well, Huang, instead of lavishing praise, asks, “What could have you done better?” Huang strives to keep internal politics out of Nvidia. When he meets with his executive team, the room is packed with many dozens of associates as he wants to disseminate information broadly rather than hold it closely among eight or ten people. He doesn’t take people aside to criticize or correct them; he publicly dresses down executives and product champions, believing that this approach is an efficiency gain for the company, outweighing the public embarrassment and humiliation of the individual. Not surprisingly, this has led to the resignations of many good managers.

In order to keep his hand firmly on the pulse of the company, Huang asks managers and product champions (he calls them “pilots in command”) to send emails with information about what they are working on and what they observe in the market. He reads around 100 of these emails every day and answers many of them. Unlike much of corporate America where executives use PowerPoint slides to present information usually accepted at face value by the participants in the meetings, Huang believes in the whiteboard. There are whiteboards throughout every Nvidia building. Huang believes that whiteboarding leads to problem-solving. In management meetings with several hundred leaders present, each general manager goes up before the group and discusses his or her business segment using the whiteboard to explain what they do, what challenges they face, and field questions from Huang and everyone else. Huang thinks that whiteboarding focuses the company on the future rather than the past. Huang also does not employ five-year plans. He says: “Strategy is not words. Strategy is action. The world is a living, breathing thing. We just plan continuously.”

Although Huang didn’t quite realize it, the first 20 years of Nvidia’s history had prepared the firm to lead the AI market over the decade to come. But it was at this point in 2013 that Huang would begin to pull every lever he could to bring about the AI-powered future. He had already taken the expensive step to make Nvidia’s hardware compatible with CUDA so that researchers and engineers could program Nvidia’s GPUs for their needs. Now his lieutenants were asked to create a software CUDA-based library which would become a necessity for AI developers. Huang was creating an Nvidia ecosystem that would eventually support the whole move to AI.

The next major step which put Nvidia in the dominant role in the AI market was the purchase of Mellanox, a company founded by several Israeli engineers. Mellanox provided high-speed networking products for data centers and supercomputers, becoming the industry leader. As its stock price had not done as well as the industry sector that it was in, an activist hedge fund named Starboard took a position in the stock and was able to name three board members. In 2018, Starboard pressured Mellanox to seek an acquirer, and a bidding war soon ensued. There were serious bidders for Mellanox — Nvidia, Intel, and Xilinx. The bidding war went on for some months, and finally Nvidia won it with a bid of $125 a share in an all-cash offer costing $6.9 billion. It turned out to be one of the most consequential transactions in the history of the semiconductor industry. In a conference call with investors and analysts following the acquisition, Huang explained why the acquisition was such a big deal. He detailed how AI applications would require tens of thousands of servers connected to one another and working in concert, and that the market-leading networking technology from Mellanox would be critical to make that possible. Huang was right. When Mellanox reported its sales in its last quarter as an independent public company in early 2020, its annual run rate was approximately $1.8 billion. Four years later as part of Nvidia, the Mellanox division was reporting annual sales of around $12 billion — this for a company that Nvidia had only paid $6.9 billion. Mellanox helped Nvidia rise to dominance in the AI space; Huang’s acquisition was sheer genius. It was another step in Nvidia’s transition from a company whose primary market was game card revenue to a hardware and software company that powered AI.

Nvidia reported its financial results for the fiscal quarter ending April 20, 2023 a month later, on May 24, 2023. They reported that sales for the fiscal quarter were $7.2 billion — down about 12% from the quarter one year earlier, and earnings per share were down significantly as well. But then came the bombshell. The analysts’ consensus for the coming fiscal quarter, July 2023, was $7.1 billion. Nvidia forecast sales of $11 billion — almost $4 billion above the consensus forecast. The analysts’ forecast was only off by 57%. In fact, the actual sales for the July quarter came in at $13.5 billion. The next day, the price of the stock climbed 24%; a year later Nvidia’s stock had appreciated over threefold. When demand for generative AI erupted in 2023, Nvidia was the only manufacturer ready to meet it.

The following week, in an address at a technology conference in Taiwan, Huang announced Nvidia’s new DGX GH200 AI supercomputer, which incorporated 256 GPUs in one system, 32 times the number of GPUs that were in the previous model. It meant enormous amounts of computing power for generative artificial intelligence applications, allowing developers to build better large learning models, AI chatbots, and data-analysis tools. In typical Huang fashion, he was pushing Nvidia to produce chip designs more quickly, transitioning from the previous two-year product cycle to a one-year cycle for its AI chips. In the 1990s, Nvidia had pushed its team to produce new graphics cards every six months — faster than the competition — and now he was pushing Nvidia to do the same for AI chips. And Nvidia has other competitive advantages: unlike competitors who produce only chips, Nvidia operates across three spaces — hardware, software, and networking. This allows Nvidia to never negotiate its prices. Huang believes that Nvidia’s products are the best, and thus it has pricing power. Generally, its prices have only gone one direction: up.

The numbers above tell the amazing story of Nvidia’s success in transitioning the company from a graphics card producing company to the backbone of the AI industry. The financial results for Nvidia’s fiscal year ending at the end of January 2025 are estimates. But assuming that they are in the ballpark, they show that Nvidia’s sales will have increased almost fivefold in the past two years. The company’s earnings per share are poised to have advanced almost 14 times since the fiscal year ending in January 2023. A remarkable story rarely equaled in the annals of Wall Street. At the end of the last fiscal quarter in October 2024, Nvidia had $38 billion in cash on hand compared to only $10 billion in debt, so the balance sheet is in good shape. The company’s return on equity is over 100 percent. As this incredible story has unfolded over the past two years, the stock has appreciated approximately eightfold. With Nvidia’s stock trading at around $130 per share as this commentary goes to press, the price/earnings ratio on 2025 earnings per share estimates of $4.20 is 31. Not inexpensive, but when judged in light of Nvidia’s estimated earnings per share growth this year of over 40%, a P/E ratio of 31 seems reasonable.

Some investors believe that Nvidia’s growth will slow or even flatten out, as the demand for AI chips will moderate and competitors will join the fray. Others think that once the appetite of companies in the Magnificent 7 stocks such as Meta Platforms, Alphabet, Amazon, and Apple has been sated, the demand for AI chips will decline. There may be some truth in this analysis. But Nvidia has great strengths. As discussed above, it leads the way in satisfying AI demand because it is the only player that can deliver the hardware, software, and networking tools that are needed. But we at BFS believe that Nvidia’s greatest strength is Jensen Huang. At the end of the day, Jensen Huang is Nvidia. He has been CEO at Nvidia for 30 years and is the fourth longest-serving CEO in America. And he is only 61 years old. He has a long runway, and he lives to work. Moreover, he has created a company with a work culture perhaps unsurpassed in this country. There are, indeed, challenges ahead for Nvidia, but we, at BFS, believe that Jensen Huang can handle them. We also believe that Nvidia, whose stock has been quite volatile over the past years, is suitable for aggressive portfolios, when it can be purchased at a reasonable price. Like Jensen Huang, we think that Nvidia is a stock for the long haul.