Elon Musk just left, and the "tough guy" Huang Renxun is coming.

After Musk left, another global business star is about to set foot on the mainland of China.

Sources revealed that Huang Renxun, CEO of NVIDIA, will visit mainland China this month. Tencent, ByteDance, Ideal Auto, BYD, Xiaomi... The list of companies on Huang Renxun's itinerary indicates that this will be a compact and not easy journey.

Huang Renxun has been busy overturning the technology industry's perception of AI these days.

On May 29th, at the COMPUTEX 2023 podium, he wore a leather jacket, black pants, and his signature smile, bringing multiple NVIDIA "heavy bombs", the most eye-catching of which is the GH200 chip, known as the "giant GPU", has been fully put into production.

In addition to hardware, NVIDIA has also launched a service for outsourcing AI models for game development. Prior to this, NVIDIA had launched cloud service products for customized large model outsourcing for language models, vision, and drug development.

After the speech, Huang Renxun rushed to the MediaTek press conference with MediaTek CEO Cai Lixing to announce that the two companies will cooperate in developing automotive cockpit chips.

This "AI Godfather" who just turned 60 this year is becoming the hottest figure in the global technology circle.

On May 25th, NVIDIA's closing price soared by 24%, rising by an Alibaba in a day; on May 30th, US Eastern Time, NVIDIA's opening price rose by more than 4%, reaching $405.93 per share, and the total market value officially exceeded 1 trillion US dollars. This means that Huang Renxun has become the first Chinese entrepreneur in history to reach a trillion-dollar market value, and NVIDIA has become the first trillion-dollar chip company in history.

Huang Renxun's confidence comes from the wave of large AI models sweeping the world led by ChatGPT. Under the global large model arms race, the demand for underlying computing power has skyrocketed, and Huang Renxun and NVIDIA, as "arms dealers", have become the biggest winners of this "hundred-model war".

Computing power, algorithms, and data constitute the three elements of the AI era, and computing power is the basic infrastructure like electricity. How many GPUs and how much computing power are critical to evaluating the success or failure of large model research and entrepreneurship. GPUs are the core components for ChatGPT training and deployment, and NVIDIA has almost no competitors in the GPU field. According to estimates, running a ChatGPT requires ten thousand NVIDIA A100 chips.

At the age of 16, he entered university, founded his own IT company at the age of 30, and the company went public at the age of 36. After 30 years of entrepreneurship, the company has remained in the trillion-dollar market value club and he still sits in the CEO position... Huang Renxun's experience is quite legendary.

But Huang Renxun and NVIDIA's success was not smooth sailing.

In his graduation speech at Taiwan University on May 27th, Huang Renxun mentioned that during the entrepreneurial period, NVIDIA stumbled several times and almost went bankrupt several times; after NVIDIA went public, the stock price once fell to only $6. What made NVIDIA repeatedly come back to life and accurately hit the trend is not a stroke of luck, but the entrepreneurial thinking that Huang Renxun has always maintained-even if the company has achieved a leading position, he has never relaxed for a moment. **

As the "most combative Chinese in Silicon Valley," there are only four things that are very important in Huang Renxun's life: work, thinking, helping others, and creating. Huang Renxun often says, "Remember, we are only 30 days away from bankruptcy," to inspire employees' combat effectiveness.

"His speech is very powerful, and he feels like he can talk non-stop from morning till night." This is the general impression of Huang Renxun by NVIDIA employees. What is commendable is that as a manager, he has to read documents until late at night every day and sleeps very little. Moreover, he has very high requirements for translation. He personally verifies the materials translated by his assistants.

According to Huang Renxun's subordinates, "Huang Renxun himself reads a lot of papers and is the one among the bosses of big companies who understands the details of new technologies the most. He manages the front-line engineers, assigns tasks to them, and checks their progress. He requires that anything that may hinder progress needs to be reported to him as soon as possible."

Huang Renxun once said, "As a company, as an entrepreneur, you must be very eager for success, more than your opponents are eager for you to die." Therefore, whether it is the metaverse, cryptocurrency, or AI, Huang Renxun has not missed any of these technology trends.

The times will not create unprepared heroes, it is better to say that the trend hit Huang Renxun.

Ambition beyond "arms dealers"

People close to Huang Renxun describe him as "a ruthless person."

As the head of a chip giant, Huang Renxun strives to do everything himself. Many of NVIDIA's cooperative affairs are negotiated by Huang Renxun himself. The origin of NVIDIA's cooperation with this round of generative AI wave was the seed planted by Huang Renxun in 2016.

At that time, OpenAI had just been established, and Huang Renxun personally went to OpenAI to exchange ideas and donated a DGX-1 with 8 P100 chips, which was NVIDIA's most powerful floating-point operation GPU at the time. Huang Renxun also mentioned this year that NVIDIA's AI supercomputer DGX is the engine behind large models, and he personally handed the world's first DGX to OpenAI.

With the help of this equipment, six years later, OpenAI launched ChatGPT and achieved innovative iteration from GPT 3.5 to 4.0. The one-year training time required was also compressed to just one month. Huang Renxun witnessed the growth of this start-up company, and NVIDIA, as a chip supplier, also boarded the high-speed development of the AIGC wave.

"The iPhone moment of AI has come." From the beginning of the year to the present, Huang Renxun has repeatedly mentioned this sentence. He believes that generative AI will reshape almost all industries. At this critical moment, NVIDIA's role is not only to provide chips as "arms dealers," but its greater ambition is the technical infrastructure of AI.

At the COMPUTEX 2023 conference, Huang Renxun brought a batch of the latest products and services related to AI. Named DGX GH200, the AI supercomputer is seen as a breakthrough technology engine that overcomes the computational limitations of ChatGPT-type products. It is expected that technology giants such as Microsoft, Meta, and Google Cloud will be granted the first access. Huang Renxun revealed that DGX GH200 integrates 256 GH200 Grace Hopper super chips together, which have been fully put into production.

Not only chips, but Huang Renxun hopes to use generative AI to reach more small and medium-sized enterprises through cloud deployment, continuously expanding the service market for AI computing power. In March of this year, NVIDIA released the AI cloud service DGX Cloud, which provides users with monthly rental services for training generative AI large models and other AI applications through cloud vendor platforms. "With just a browser, NVIDIA's AI computing power can be accessed by every company."

Huang Renxun also demonstrated several new generative AI applications. In addition, NVIDIA will provide a game-customized AI model foundry service, Avatar Cloud Engine (ACE) for Games. In the hot field of automobiles, NVIDIA has also made some progress.

Huang Renxun said: "In the next decade, our industry will use new AI computers to replace traditional computers worth trillions of dollars." In the future, companies and talents without AI expertise will fall behind. Because new AI technologies will change the industry landscape and affect every aspect of a company's work.

Huang Renxun's entrepreneurial story

Huang Renxun's entrepreneurship illustrates how a company actively cultivates a niche market, perseveres in growing, and leverages the trend of the times. His growth with NVIDIA is a perfect footnote to the Chinese tradition of "timing" and "momentum".

In 1993, Huang Renxun, Chris Malachowsky (currently NVIDIA's senior technology officer), and Curtis Priem (retired in 2003) co-founded NVIDIA. They targeted the field of graphics processing.

This field was not the mainstream of the market at that time.

In the 1980s and 1990s, the dominant Intel and AMD were competing in the direction of personal computer microprocessors, using x86 systems. Intel completely dominated the CPU-centric microprocessor industry. However, Intel's chip empire did not monopolize the graphics processing industry, which gave NVIDIA the opportunity to cultivate this market and survive.

Just as innovation on the edge of the market, NVIDIA's birthplace also has a metaphor. It did not originate in the Palo Alto cafe in Silicon Valley, but in a Denny's fast food restaurant in San Jose.

This is related to the background of the times. In the 1990s, the "fab-less" revolution swept the semiconductor industry, especially the rise of TSMC, which intensified the trend of offshore outsourcing. Fortunately, NVIDIA did not need to establish its own wafer factory, otherwise it would not be able to raise the necessary funds in the start-up stage. It should be noted that building a wafer factory at the end of the 1990s required 100 million US dollars. NVIDIA's initial startup capital was only $40,000.

"I met with Mr. Zhang Zhongmou of TSMC and began a 25-year partnership," Huang Renxun recalled in a speech at National Taiwan University. From its inception to today, NVIDIA's chips have been mainly manufactured by TSMC. If NVIDIA were to manage its own wafer factory, it might not have enough resources to build a software ecosystem.

In 2006, NVIDIA realized that high-speed parallel computing could be used for users beyond computer graphics and released CUDA (Unified Computing Device Architecture) software, which allows GPUs to be programmed in standard programming languages without any graphics references.

According to NVIDIA's estimate in 2017, Huang Renxun spent at least $10 billion on this software engineering project to make programmers, not just graphics experts, use NVIDIA's chips. Huang Renxun donated the CUDA software, but the software is only applicable to NVIDIA's chips.

A chip industry insider said that NVIDIA's strongest asset is not its hardware, such as GPUs, but its ecosystem. Every vendor trying to enter the graphics computing market must use its software, and to achieve optimal performance, vendors must purchase its chips.

But this process is also quite difficult. Huang Renxun mentioned three entrepreneurial stories in his speech at National Taiwan University, about how to deal with mistakes, how to learn to persevere, and when to give up. These qualities shape NVIDIA as a company.

The first story happened in the early days of the company's establishment. At that time, NVIDIA launched a 3D graphics application for personal computer games and won a contract with SEGA to build game consoles. However, after a year of development, Huang Renxun realized that NVIDIA's architecture was a wrong strategy. If SEGA withdrew from the partnership and did not pay, NVIDIA would face bankruptcy. Huang Renxun convinced SEGA to pay the full cost and developed a new product.

"We face mistakes honestly, humbly seek help, and save NVIDIA's existence. These qualities are the hardest to develop for the smartest and most successful people," Huang Renxun said.

The second story happened in 2007. NVIDIA announced the CUDA GPU acceleration computing technology, but the cost of CUDA was very high, and NVIDIA's profits suffered huge blows for many years, with a market value of only around $1 billion. Years of poor performance made shareholders skeptical of CUDA and hoped to focus on improving profitability. But NVIDIA persisted.

Huang Renxun said that we invented CUDA for AI, and this journey forged our character, enduring pain and suffering, which is a necessary pain on the road to pursuing our vision.

The third story happened in 2010. NVIDIA withdrew from the huge mobile phone market and created a robot market with an unknown market size. Now, NVIDIA has a business of billions of dollars in autonomous driving and robot technology.

"'Retreat' is not easy for smart and successful people. However, strategic retreat, sacrifice, and deciding what to give up are the core of success, a very critical core," Huang Renxun said. Silicon Valley entrepreneurs and scientists have their own theories, such as Gordon Moore, the founder of Intel, and CEO Andy Grove, who are famous for "Moore's Law" and "disruptive innovation," respectively. Jerry Sanders, founder and CEO of AMD, is the brightest salesman in Silicon Valley.

Different from the intelligence of Silicon Valley scientists, Huang Renxun is an engineer, but he is more of a typical business talent. Huang Renxun carries the shadow of "Grove", but he is not overturning an industry, but creating an industry. Different from the rationality and technical personality of the Western style, Huang Renxun is more adventurous and more willing to take risks.

Previously, when Huang Renxun worked at LSI Logic, he worked in the design department for two years and then requested to transfer to the sales department. "Transferring from the engineering department to the sales department was the best career choice I ever made. I gradually learned how to combine product design and development with the market. Because I realized that consumers don't care where you graduated from business school, they only care about one thing, what benefits your product can bring to them," he summarized.

The Key Chinese Market

In 2013, at the Xiaomi 3 press conference, Huang Renxun flew to Beijing specifically to support Lei Jun. He introduced Nvidia in poor Chinese because Lei Jun asked him to speak Chinese. Huang Renxun was born in Taiwan, China, but went to the United States to live and study at the age of 9.

Huang Renxun introduced in Chinese at the time: "We invented the GPU, our GPU is the fastest and the best." Someone shouted below, "There is no one else."

A top domestic scientist recalled that around 2011 or 2012, Huang Renxun himself didn't know what GPUs could be used for. "Where did he have foresight? If he said it was artificial intelligence at the time, it was all a scam," because at that time in China, everyone's energy was focused on making mobile phones and on mobile Internet. But Huang Renxun used GPU technology on mobile phones. He made chips for mobile phone manufacturers all over the world when other chip companies were unwilling to do so.

China's booming mobile Internet and mobile phone market became Nvidia's cash cow for a period of time, supporting Nvidia through its most difficult times. An industry insider in the autonomous driving industry also said that Nvidia's largest market for in-vehicle intelligent chips is in China, which also provides the richest scene exploration.

Huang Renxun and Nvidia are inseparable from the Chinese market.

On May 24th, Huang Renxun once again stated that export control measures on chips could cause "huge damage" to the US technology industry. The export control order has tied Nvidia's "hands behind its back" and prevented it from selling advanced chips in one of its largest markets. Huang Renxun emphasized that China accounts for about one-third of the US technology industry market and cannot be replaced.

Nvidia's financial report for the 2023 fiscal year (February 2022 to January 2023) shows that Nvidia's revenue in the Chinese market accounts for 47% of its overall global market revenue. Huang Renxun certainly does not want Nvidia's most important Chinese market to be taken away. Currently, Nvidia's A100 and the latest H100 chips cannot enter the Chinese market and can only enter the Chinese market in the form of "A800" and "H800" products with some performance cut off.

Chinese tech giants BAT and a group of startups are launching a new round of competition in the AI field, and their demand for GPU chips has risen to hundreds of thousands. Some startups have to rent A100 resources from large companies to develop their own large models.

The domestic large model craze is causing a computing power gap, and where there is market demand, there will be commercial companies entering.

A domestic server manufacturer said that the rise of AIGC Inspur is a big opportunity for Nvidia, but at the same time, the followers behind cannot be underestimated. Everyone has certain technical strength and can find corresponding segmentation scenarios. Whether it is a startup or a new product invested by a large factory, there is an opportunity. This is a very dialectical and open era.

On May 30th, Huang Renxun said at an event in Taipei, "Don't underestimate the chasing ability of Chinese companies. "We must run very fast ourselves." He said, "China's investment in this area... is quite large, so you cannot underestimate them."

Although Nvidia officials did not have information on the executive's itinerary in response to "China Entrepreneur" magazine, Huang Renxun came to China this summer and met with TSMC, Foxconn, Tencent, ByteDance, Xiaomi and other companies across the strait, which shows his sufficient attention to the Chinese market.

The technological iteration of chips and market demand are mutually causal. Where the market is, the latest technological advances in chip demand will emerge. In the era of AI large models, Nvidia has once again entered a period of high-speed growth.

However, when a commercial empire is at its peak, it is also the most dangerous moment. Intel dominated the personal computer microprocessor and CPU, but missed the rise of mobile internet and smartphones. Having seen the rise and fall of Intel, Huang Renxun should be concerned.

But Huang Renxun has never been afraid of technological trends, but enjoys them to the fullest. At the age of 60, he said he has no plans to retire yet, "leading Nvidia for another 30 to 40 years, until around 90 years old," and will continue to work in the form of robots.

Huang Renxun's first half of his life has never stopped. As he said, "Whether it's running for food or not being treated as food by others, you often don't know which situation you are in, but no matter what, you must keep running."