Huang Renxun and NVIDIA, Part 01: Whose Wife Agreed to the Startup First?

Hi everyone, welcome to this episode of Cold-Knowledge Blind Box. NVIDIA has recently been allowed to resume sales of the H20 graphics card, and CEO Huang Renxun has come to China for business talks. As the head of the world's most valuable company, he is someone many people are curious about. So we're starting a special series on Huang Renxun and NVIDIA.

Okay, let's get started!

Huang Renxun is a source of Chinese pride, so surely he had a good elite education? That's half right. His parents made a serious mistake when choosing a school for him.

At the time, Huang Renxun's parents were still in Thailand processing their immigration papers. Because the political situation there was unstable, they had to leave the young Huang Renxun with relatives in the United States first. This was the 1970s, before the internet, so information moved slowly. The relatives were also recent immigrants who didn't speak English, and his parents were on the other side of the ocean.

As a result, the family wrongly assumed that every boarding school in America was the kind of elite aristocratic boarding school they had imagined. They spent everything they had to send Huang Renxun to a school called Oneida Friends School. In reality, it was a mountain reform school that used Christian principles to reform young offenders with a history of violence.

A Hard School Life

Huang Renxun and his older brother were stunned as soon as they arrived. Huang Renxun was only nine years old, and his roommate was already 17, covered in tattoos and knife scars. By then it was too late to withdraw. In that era, bullying was routine, and Asian kids were especially targeted.

Apart from classes, the two boys had to do labor as part of their "reform":

His older brother worked on the farm, growing tobacco
Huang Renxun was responsible for cleaning the toilets in a three-story dormitory building

Turning Enemies into Friends

His roommate was a delinquent teenager, and there was no way to change rooms. Since he couldn't beat his opponent, he had no choice but to join them. When Huang Renxun was very young, his mother picked ten random words from a dictionary every day and made him memorize them. Nine-year-old Huang Renxun used that skill to teach the 17-year-old American delinquent English words.

The two quickly became good friends. The older boy looked out for Huang Renxun, taught him weightlifting and fitness, and hit back for him whenever someone picked a fight.

The Impact of the Reform School Years

Two years later, his parents successfully immigrated to the United States from Thailand, rescued him from the reform school, and transferred him to a proper public school.

Those two years at the reform school played a major role in shaping his character:

He admits that this period forged his tenacious, street-fighter personality
It is also why NVIDIA has nearly gone bankrupt several times and still always pulled through

To repay his old school, Huang Renxun wrote a $2 million check in 2019 to build a new dormitory building there.

The Toilet-Cleaning Skill

Huang Renxun holds a Silicon Valley record that nobody has broken. He has admitted that he is the Silicon Valley CEO who has cleaned the most toilets, and the one who is best at it, thanks to the toilet-cleaning skills he built at the reform school.

Later, while working a part-time job in high school, he went to Denny's, a cheap American restaurant chain, and continued cleaning toilets. Denny's was a 24-hour diner where a meal cost about $10, and many customers were North American long-haul truck drivers working the night shift, so you can imagine how clean the restrooms were.

While working at Denny's, he learned:

How to balance quality and efficiency
How to cope with chaos in the kitchen
How to communicate with customers

In 1993, Huang Renxun and two partners sketched out NVIDIA's startup plan on a restaurant placemat at Denny's. To commemorate the moment, Denny's later launched an "NVIDIA Breakfast."

College Years

Studies and Romance

An early bloomer, Huang Renxun graduated from high school at 16 and entered Oregon State University to study electrical engineering. There were almost no women in engineering. In his class of 250, only three were women, and Huang Renxun managed to catch the attention of his future wife, Lori.

With 247 men chasing three women, that's an elimination rate of 98.8%. Becoming friends with a delinquent older boy at nine, and winning over a girlfriend in college, both show that he had outstanding interpersonal skills from a very young age.

The two married right after graduating and have now been married for 41 years, with two children. A first love that lasted a lifetime suggests Huang Renxun put all of his energy into his career.

Job Hunting and Career

After graduating, Huang Renxun's job hunt didn't go smoothly at first, and Texas Instruments turned him down. Interestingly, when he interviewed at TI, another Taiwanese man, Zhang Zhongmou, had already risen to global vice president there. By the time Huang Renxun sent in his résumé, Zhang Zhongmou had just left TI to return to Taiwan and found TSMC.

If Zhang Zhongmou had gone back to Taiwan a few years later, Huang Renxun might well have followed him and become a founding member of TSMC's startup team.

Although Huang Renxun didn't get into TI, he was hired by AMD. He worked at AMD during the day and studied for a master's degree at Stanford on weekends. Even so busy, he and his wife still found time to have a son and a daughter.

Because his time was limited, it took him eight years in total to finish his master's. Most people would have given up long before. Huang Renxun said: "For some things I want to see change right away, but for others I have infinite patience, and I'll keep at them."

Career Turning Point

From AMD to LSI

Although Huang Renxun stuck it out for eight years to earn his master's degree, he was already thinking about leaving AMD just a year after he started. At the time, AMD's chips were still designed by engineers by hand, using calculators, and Huang Renxun saw clearly that this approach had no future.

By contrast, another company, LSI, had developed a computer-aided design (CAD) system that let customers log in remotely and design chips themselves. So Huang Renxun decided to leave and join LSI.

If he had stayed at AMD for eight years, the way he did for his master's, he might now be an ordinary Chinese engineer whose greatest achievement is a nice house in Silicon Valley.

The Wisdom of Persistence and Letting Go

How do you tell which things to stick with and which to give up? Huang Renxun's explanation: "Greatness has nothing to do with being smart. It depends on character."

The major choices at life's crossroads ultimately rest on the character he built at nine, cleaning toilets and facing bullying. The two years at the reform school taught him to be a tenacious street fighter, not to accept the status quo and passively endure.

In his later biography, Huang Renxun wrote: "I hope young people can go through enough pain and hardship."

How NVIDIA's Three Founders Met

Priem's Path as a Genius

After joining LSI, Huang Renxun was sent to Sun Microsystems to work on a project, and there he met the other two NVIDIA founders: Curtis Priem and Chris Malachowsky.

Curtis Priem was a computer genius. When he graduated from high school, he was admitted to three universities at once, one of them MIT. He decisively turned down MIT, ranked among the world's top three, and chose RPI, a technical school ranked around 70th in the United States.

The reason was that the school had just bought one of the most advanced IBM 3033 mainframes of the time, which gave him a chance to program on it. After he enrolled, he spent so much time on the machine that its heat could have warmed the entire building through the winter.

The Timing

When Priem was in college, Steve Jobs released the Apple II. The machine proved that computers didn't have to live in school computer rooms or corporate offices. They could sit on every ordinary family's desk as a consumer product. That also meant ordinary users would soon demand better graphics.

Priem's first job was designing the most advanced graphics cards. In 1984, the card he designed could deliver 640×480 resolution, a big step up from the previous generation's 640×200, and the number of colors jumped from 16 to 256.

Setbacks at Work

But like many technical geniuses, Priem had terrible luck at work. Although he could design the most advanced graphics cards, he kept landing at bad companies that gave him no stock options, while handing him plenty of free exposure to office politics and management squabbles.

Eventually, after many twists and turns, Priem joined Sun Microsystems.

The Birth of the GX Graphics Card

A Secret Project

At Sun Microsystems, Priem met his first champion, an executive who had a secret project of his own. He planned to develop advanced graphics hardware behind the backs of Sun's senior management. This meant moving part of the graphics processing work off the CPU and onto a piece of hardware that would later be called a graphics card.

A Fight Over Technical Direction

Such a bold idea sparked a battle over technical direction inside the company. Fights over technical roadmaps at tech companies were as intense as ideological struggles in a communist revolution.

Sun's top leadership wanted to focus on improving CPU performance, which they saw as the core business of server computers. So the project had to be carried out in secret.

Building the Team

Faced with such a difficult technical challenge and an enormous amount of freedom, Priem said just one thing to his boss: "Give me a team."

So the boss recruited a brilliant engineer from HP, who would later become another NVIDIA co-founder: Chris Malachowsky. He would be responsible for turning Priem's brilliant ideas into concrete chip designs.

To make sure the secret project moved forward smoothly, LSI, an important partner of Sun Microsystems, also sent a rising star of the company, Huang Renxun, to join the project's integration and manufacturing work.

At this point, the destinies of NVIDIA's three founders had crossed.

Getting the Team to Gel

Hi everyone, welcome to Cold-Knowledge Blind Box. Picking up where we left off, the three NVIDIA founders' destinies had crossed. In this episode, I'll keep going with how NVIDIA was founded and its early startup story.

Priem was responsible for the blueprint of the GPU, and he is American. Malachowsky was the team's engineer, responsible for turning the GPU from a blueprint into a real product, and he came from the former Soviet Union. Huang Renxun handled project management.

A typical scene was Priem and Malachowsky staging a Silicon Valley version of the US-Soviet standoff in the office. After every loud argument, Priem would storm off in a sulk, and the engineers on the team would blink nervously and ask, "Are we about to break up?"

Huang Renxun's answer: "Intense arguments may be uncomfortable, but to get the very best ideas, you have to embrace conflict."

This later became NVIDIA's corporate culture.

The Groundbreaking GX Card

The Product

In 1988, this secret team produced a groundbreaking product: the GX graphics card. How impressive was it?

To show off the card's hardware power, Priem made a flight-combat game in his spare time, featuring the F/A-18 Hornet. It is said to be the first flight simulation game in human history. Priem even bought real satellite data to build the game's battlefield terrain, and the game supported LAN multiplayer.

Comparing the Specs

What resolution did this air-combat game run at? 1280×1024.

Six years later, the famous Taiwanese RPG Chinese Paladin, from Softstar, reached a maximum resolution of only 320×200. At the time, only the GX card could fully support the outrageous graphics demands of this air-combat game. It was pure showing off.

Commercial Success

Faced with such an outstanding product, Sun's senior management changed its hostile attitude toward graphics cards. Soon the card was priced at $2,000 and sold as an optional upgrade kit for servers.

The flight-combat game was sent to every Sun sales office. Salespeople demonstrated the servers' excellent graphics performance to corporate customers and crushed their competitors.

The upgrade kit sold like hotcakes and quickly became the standard default configuration for Sun servers. LSI, where Huang Renxun worked, saw its revenue jump from $262 million to $656 million.

Huang Renxun was promoted to department director for this successful project.

The Trigger for Starting a Company

Infighting at the Company

But after these young people brought Sun commercial success, the company quickly sank into endless infighting and political struggles. For two years it released no new products and lived off the old GX card.

Priem, who had contributed enormously, was pushed out inside the company and was even moved to tears at one point. (He's a tech nerd, so everyone understands.)

The Startup Idea

Malachowsky, his good buddy at the company, had had enough, and suggested that the two of them leave Sun together to start a company. But these two tech nerds had no business sense. At that point, they both thought of the same person at once: Huang Renxun, whom they had worked with on the GX project two years earlier.

Timing Analysis

In hindsight, the timing of their startup was ideal. In 1992, personal computers got two revolutionary developments:

The PCI bus
The Windows 3.1 operating system

The spread of personal PCs thus entered an explosive phase.

All three of them knew the startup would be about graphics for personal PCs, but the specific business direction was still a blank.

The Business Plan at Denny's

Drafting the Plan

At the end of 1992, the three began writing their business plan at a Denny's in East San Jose. Huang Renxun's usual order there was a sandwich called the "Super Bird," with turkey, melted Swiss cheese, tomato, and his favorite, bacon.

Huang Renxun's Hesitation

At first Huang Renxun seemed a bit undecided, because his job was going smoothly and he had just been promoted to director. The other two partners, however, were in a tough spot. Anyone would think: "Why should I give up my bright future to start a company with you two?"

Fortunately, Huang Renxun was a gamer himself, and Priem and Malachowsky soon convinced him: if they developed graphics cards for gamers, the company could bring in more than $50 million in revenue in a very short time.

The Wives' Showdown

The next question was which of the three would quit first. This led to a funny episode: the Wives' Showdown. Each of their wives wanted the other men to quit first.

In the end, the one who broke the deadlock was Priem, who was doing the worst at work and was about to be fired anyway.

Quitting to Start the Company

Leaving in Stages

December 1992: Priem officially resigned. (Not only Huang Renxun, even his good buddy Malachowsky hadn't left yet.)
March 1993: Malachowsky finished the GX card upgrade and also resigned. His boss, the one who had secretly had him develop the GX card, gave him a parting gift: permission to take one Sun server with him to his new company. A few days before he left, the resourceful Malachowsky used company funds to upgrade the machine to the largest memory, the biggest hard drive, and the largest monitor available.
June 1993: Huang Renxun also officially resigned. He had just turned 30 that year.

Founding NVIDIA

The three set off together on a great voyage. When they registered the company, each of them put in $200 and bought a third of NVIDIA. The company's latest market value is around $3 trillion.

Priem lived in the suburbs, where the house had more room, so he cleared out a room and moved Malachowsky's server in. NVIDIA officially opened for business right in his house.

The Early Team

The three had already built a bit of a name for themselves in Silicon Valley, so soon after they left, a group of talented people gathered around them. Twelve employees on the early team worked without any salary.

The company's most urgent task was to build a first product to show, then take the demo to venture capital firms to raise money and pay the first salaries.

The Sequoia Capital Pitch

Preparing the Pitch

If there are startup founders watching who are also trying to raise money, you may be wondering: when pitching to VCs, is the product demo more important, or is the founding team's ability to improvise in the moment?

The answer: neither matters.

A Train Wreck

The three founders' first major pitch to a VC was a huge train wreck. The person interviewing them was Don Valentine, the boss of Sequoia Capital, now the world's largest venture capital firm, which had also been an early investor in Google and Apple.

For the pitch, Huang Renxun brought a modified GX card as a concept demo. Although the card had been out for four years, its performance still crushed any Windows graphics card on the market. They even brought an early VR headset to demonstrate a game.

Valentine's Doubts

But what the NVIDIA team didn't know was that Don Valentine most disliked flashy product demos. He had seen too many products with impressive demos end up losing badly in real commercial competition.

After the demo, Valentine asked: "Are you a game console company, a computer graphics company, or an audio company?"

Priem was stumped on the spot, then delivered a long speech full of technical detail to prove that NVIDIA was the sum of all of the above. Most of what he said was correct. NVIDIA later did achieve "one card, many uses," with a single graphics card serving game consoles, computer graphics and audio processing all at once.

What he really wanted to say was this: if one chip can handle many tasks, the cost of hardware development will drop dramatically.

Valentine was unhappy with an answer that sounded like evasion. He said: "You have to choose one, or you'll fail, because you don't know who you are."

The three young men were speechless, and the atmosphere in the room was extremely awkward.

A Visionary Technical Idea

Valentine went on: "What will NVIDIA look like ten years from now?"

The tech nerd Priem answered again: "We will have a unified shader architecture." What he actually meant was that future graphics cards would no longer be just graphics accelerators. They would become high-performance parallel platforms for general-purpose computing.

Today's GPUs, through CUDA and the unified shader architecture, can work together with CPU memory to efficiently handle the large-scale matrix computations that AI needs. But in the early 1990s, the idea was far too advanced. Valentine looked completely baffled, and even Huang Renxun and Malachowsky couldn't follow what he was talking about.

Choosing a Foundry

Then Valentine asked which company they planned to have make the graphics chips. Huang Renxun said they planned to use the European chipmaker STMicroelectronics, though at that time the company itself was nearly bankrupt. Valentine recommended TSMC in Taiwan instead.

An Unexpected Investment

The Pitch Failed

After the pitch, all three were deeply dejected. Huang Renxun, as CEO, took responsibility for the failure on his own shoulders. Malachowsky later described the pitch as being like a courtroom hearing before a judge.

The Real Reasons They Got the Investment

Unexpectedly, though, Sequoia agreed to invest very quickly.

First reason: The CEO of LSI, where Huang Renxun had worked before, called Valentine and gave a highly favorable assessment of Huang Renxun as a person. Every venture capitalist knows that a startup team's reputation in the industry matters more than how it performs in a pitch.

Second, and most important, reason: Sequoia had just invested in the famous EA and seen it go public successfully, earning a huge windfall, so the firm was very bullish on the future of the gaming market. It had also invested in one of NVIDIA's main rivals, and Huang Renxun's team believed they could beat S3, which was strong in 2D graphics, by building 3D graphics cards. Investing in NVIDIA was effectively Sequoia hedging its bets on both sides, so it couldn't lose.

Closing the Round

In June 1993, NVIDIA raised a total of $2 million in Series A funding from Sequoia Capital and another venture firm, Sutter Hill.

After the contract was signed, Valentine said to Huang Renxun: "Investing in you went against my instincts, but your old boss, the CEO of LSI, suggested I give you some money to try. If you lose all my money, I will definitely find someone to take you out. Your business plan is badly written, but your reputation in the industry will get ahead of you and give the VCs a good first impression."

Wrap-Up and Looking Ahead

This shows why Sequoia Capital is so formidable: its boss was able to make many counterintuitive, even counter-human investments.

Next episode preview: I'll cover the story of how NVIDIA, after getting the money, competed with 3dfx and its Voodoo graphics card.

If you're interested in the book The Nvidia Way, don't forget to like the video and add it to your personal collection. Now let's bring this vanished text back with technology. See you next time!