On October 18, 2023, the U.S. Department of Commerce announced a further contraction of chip exports to China and clearly stipulated the hardware performance indicators of restricted chips.
The most prohibitive ban in history has attracted attention since its introduction.
It is worth mentioning that NVIDIA's highest-performance chips A100 and H100 are on the US blacklist, and what is even more ironic is that Huang Jenxun, the founder of NVIDIA, is still a yellow-skinned Chinese-American.
What is the origin of this Silicon Valley tycoon who often appears in front of people in a leather jacket and black pants with a trademark smile on his face?
Huang was born in Taipei in 1963.
His father was a chemical engineer and his mother was an elementary school teacher. Compared to the highlights of his peak, Huang's background seems unremarkable, and even full of a bit of gloomy undertones, which is not an exaggeration to describe as displaced.
In the 70s, the situation in Taiwan was turbulent, and the family had to move to Thailand. However, the war in Vietnam was in full swing, and the flames of war gradually spread to Thailand.
For the sake of his son Jensen Huang's future, Huang's father gritted his teeth and sent Huang, who had just turned 9, on a plane to the United States alone.
But relatives in the United States were also too busy to take care of themselves, and his uncle, who was not good at English, misread the admissions information and sent Huang, who was supposed to go to a preparatory school, to a boarding school in Kentucky.
The boarding school mainly accepts troubled teenagers from the local area, so it has long been common for students to carry daggers with them, and teachers are not surprised by students who are covered in tattoos.
During this time, Huang also shared a thorn with a prickly head who had just come out of the juvenile detention center, and was inevitably punched and kicked by the school bully, and he was arranged by the school to clean the school's men's restroom after school.
As an ethnic minority, Huang was not depressed in such a hostile environment.
On the contrary, he slowly accepted and adapted to the environment, and became good buddies with the other party by helping the thorn head make up for the lessons, and found his own way to survive.
And how to find a way out of the chaos became Huang's biggest gain during this period, and it also made him develop a character that is not afraid of difficulties.
Fortunately, Huang's parents immigrated to the United States two years later, ending Huang's poor boarding life.
In formal schools, Huang has thoroughly demonstrated the genetic talent of Chinese.
At the age of 15, he not only excelled in his grades and self-taught ability, but also excelled in table tennis, finishing third in the United States in doubles in the professional competition.
But Huang didn't join the movement like Gu did.
Instead, Huang slowly fell in love with technology and was determined to pay for his passion.
After the high school entrance examination, Huang took a break from his interests and entered Oregon State University to study electronics.
Later, he successfully entered Stanford University for further studies and successfully obtained a master's degree.
His deep cultivation in the electronics industry also laid a solid foundation for him to create NVIDIA in the future. After graduating, Huang chose to work as a chip designer at AMD. Two years later, eager to continue growing, Huang went to work as a salesperson at another chip company, LSI Logic, where he stayed for eight years.
Huang has benefited from 10 years of experience in chip companies, and his comprehensive experience from development to sales has given him a deep understanding of the intrinsic needs of chip products.
With his knowledge and love for the chip market, the 30-year-old decided to start his own company with his best friend, taking the words "Envy" and "Next Version" and naming the new company Nvidia.
But NVIDIA, which was just starting out, was splashed with cold water by experts.
At that time, CPUs were the mainstream of the market, and graphics processing units (GPUs) seemed to have no application prospects other than making graphics cards for otaku to play games.
Not to mention the mixed market. More than 30 companies, large and small, want a piece of the GPU industry. In the face of experts' efforts to dissuade him, Huang was determined to enter the game.
Of course, Huang Jenxun is not "stunned", and he is accustomed to taking the sword in an uncontrollable environment, and he vaguely smells a glimmer of business opportunity. After all, in the blue ocean realm, it is more likely to discover scenery that others have not noticed.
The maverick Huang simply didn't do anything, leaving aside the mainstream triangle drawing and rendering technology, but instead picking up the square imaging technology that no one cared about, and risking his bet on the versatility of the chip.
And just like that, NVIDIA's first product, the NV1, came out shortly after.
It is a graphics card that integrates 2D display and 3D acceleration functions, although the chip is full of technology, but the cost is very high, which makes many consumers prohibitive.
Even more lethal is Huang's insistence on using square drawing technology, which is simply incompatible with the triangular drawing technology on the market.
Seeing that the company's start-up funding had bottomed out, but the product was not profitable, Huang had to drastically cut its workforce, reducing the company's size by two-thirds.
However, at the darkest moment of NVIDIA, Japanese game console manufacturer Sega threw an olive branch to Huang, and an order worth $7 million really alleviated NVIDIA's urgent need.
It turned out that at a time when Huang was in a hurry, the Sega Saturn console was looking for a chip with both 2D and 3D performance, but fortunately the console didn't need to think about compatibility. So NVIDIA, as the only manufacturer, just catered to Sega's needs, and the two sides hit it off.
As a result of the successful completion of the first collaboration, Sega decided to allocate $7 million and send experts to fund Nvidia's NV2 chip development, so that the final chip could be used in its own Dreamcast. It's just that because the square drawing technology still has flaws, Sega hopes that NVIDIA can use the mainstream triangle drawing rendering technology.
However, NVIDIA firmly believed in its instincts and insisted on using square drawing, which led to the failure of NV2 development and directly led to the difficult birth of Dreamcast. If Sega insists that Nvidia return the money, the company Huang has created will have to face bankruptcy.
At the critical moment, Huang sincerely negotiated with Sega and admitted that the development had failed. I don't know if it was Huang's honesty that moved Sega, but the $7 million research and development fee was not returned by Sega.
It has to be said that Huang's audacity inadvertently saved the company. Originally, due to the failure of research and development, Nvidia was already in danger, and it relied on Huang Jenxun to turn the tide and live for another three months.
But the failure also made Huang more focused on technology options, and he began to actively absorb new technologies in the industry.
Nvidia employees revealed that Huang often consults many papers and participates in supercomputing conferences to learn about industry trends. He also communicates directly with front-line engineers, explaining tasks to them and checking progress.
Huang became the most tech-savvy CEO of a chip company. Even when he gave an impromptu speech, Huang was able to explain the technical details clearly, even the more hardcore industry dynamics.
You know, this is an achievement that few CEOs of Silicon Valley companies have been able to achieve.
In 1997, after the failure of Nvidia NV2, the era of 3D graphics had crept in.
With the precise control of technology, NVIDIA took advantage of the situation to launch the third-generation product Riva128, which was favored by the market by relying on the performance of far more than 4 times that of competing products, and sold 1 million pieces in four months, and NVIDIA completely stood up.
In 1999, NVIDIA launched the world's first GPU, the GeForce 256 graphics card.
At first, Nvidia simply built the best 3D graphics chips for gamers, but over time Huang discovered that GPUs could be used to speed up more complex and massive calculations.
The computing power market behind GPUs is in the dawn stage, and the prospects are not very clear. But Huang decided to gamble on everything, on this uncharted territory.
Huang appointed David Kirk as chief scientist for a project called CUDA, which aims to create a general-purpose parallel computing template that maximizes GPUs' computing power.
In Huang's eyes, GPUs are no longer just graphics chips, but keys to unlocking the future. However, it was still a decade before the explosion of AI.
Of course, Huang's desperate gamble scared shareholders out.
Reminiscent of Nvidia, which was almost planted on NV2 at the beginning, is now betting 500 million on GPUs every year on the basis of annual revenue of less than $3 billion. If the development fails, Nvidia will never recover.
At that time, Huang was under a lot of pressure, and when the company's profitability was dragged down by the CUDA platform, countless investors couldn't sit still, and they approached Huang and persuaded him to change his course.
But Huang resolutely refused to waver when he made up his mind, and if he recognized one thing in his eyes, he would never turn back easily without hitting the south wall.
And a lot of times, the South Wall isn't really there to stop you, it's just to prove how eager you really want to get there.
Luckily, Huang got the gamble right this time.
In 2006, the CUDA architecture was officially launched. Two years later, AI professor Andrew Ng built the first deep learning model based on CUDA, making GPUs emerge as computing power rather than graphics cards for the first time.
A later event completely changed the industry's perception of GPUs.
In 2012, Ng led Google Brain, which used artificial intelligence technology to identify a cat from 10 million images. He first used 16,000 CPUs, assembled 1,000 computers, and spent $1 million to accomplish his goal.
However, research projects are often time-constrained and heavy-tasked, and it is impossible to invest such huge manpower and material resources every time. So, Ng decided to try his hand at GPUs.
Shockingly, compared to the cumbersomeness of the CPU, this time Ng did the same thing with only 16 computers and 64 GPUs.
What excites Ng even more is that a new computing revolution is sweeping in, subverting the entire industry.
After all, since Intel pioneered the X86 architecture in 1978, CPUs have dominated the chip industry for nearly half a century. This kind of processor that is good at logical operations, but after the birth of artificial intelligence, it is unable to do big data operations.
NVIDIA's "parallel computing" cleverly solves the problem of thousands of threads processing at the same time, which is naturally suitable for "large and simple" operations.
To put it simply, a CPU is like a doctoral student, who is proficient in complex algorithms, but can only bury its head in solving one problem at a time, while GPU is like a group of primary school students who can skillfully apply four operations, and when faced with a large number of simple computing tasks, a group of primary school students can easily beat a doctoral student.
Therefore, in the era of big data, the sudden emergence of GPUs has directly crushed CPUs that are proficient in traditional computing, and they have become the brightest "cubs" in the AI era.
The market has made the final choice with money.
In July 2020, Nvidia surpassed Intel with a market value of $250 billion to achieve a two-level reversal, and by March 2023, Nvidia's market value will exceed $500 billion, while Intel has less than $110 billion.
On May 30, 2023, in just a few hours, Nvidia's stock soared by $200 billion, directly rising to one AMD and nearly two Intels, and its market value once exceeded $1 trillion, becoming the world's second largest technology company.
As a result, Huang's worth soared to $44 billion (313.28 billion yuan), a surge of 214.4 billion yuan from last year!
From the once precarious game graphics card company to the king of computing power in the era of big data, Nvidia's accumulation is inseparable from Huang's persistence in technology and courage to be the first.
In 2016, NVIDIA designed its first supercomputer, DGX-1, with a newly developed GPU product, the P100.
Four months later, Huang hand-delivered a DGX-1 to the newly founded artificial intelligence agency OpenAI, one of the founders of Elon Musk, known as the "Iron Man of Silicon Valley."
With NVIDIA's DGX system, OpenAI has trained the hot ChatGPT at the moment.
As computing power continues to break through the limits, NVIDIA has successively launched the more powerful A100 and H100, ushering in a qualitative leap in AI.
Nowadays, NVIDIA has reached the first choice for AI computing power, whether in terms of performance or versatility, as long as it does AI, it cannot bypass NVIDIA's chips.
Eighty or ninety percent of the assets of an AI startup are also spent on buying computing power, and chip prices are rising.
But even so, Nvidia's chips have long been in short supply, and even OpenAI CEO Sam has to plead with people not to overuse ChatGPT, because even the "landlord's house" does not have enough GPUs to optimize ChatGPT.
Today, NVIDIA's business extends to many fields.
In the field of graphics processing, Nvidia graphics cards are still in a monopoly position and provide technical support for virtual reality. In the automotive field, NVIDIA's in-vehicle computers also provide solutions and product foundations for autonomous driving.
Of course, even with the highest market capitalization chip maker in the United States, Huang has always remained sober, and he understands that only by constantly running can he stay out of the crisis. After all, he has seen too much about the origin and destruction of the growth myth.
Just like the price of graphics cards skyrocketed in 2017 when Bitcoin caught fire, Nvidia chips were also hot in the digital currency market at that time, but just a year later, with the collapse of cryptocurrencies, the price and demand of GPUs also ushered in a rollercoaster drop.
The rise of the metaverse in 2021 made Nvidia's stock price rise again, but unfortunately the market response was flat soon after.
Not to mention the backstabbing from competitors, after the rise of the old company AMD in the change of leadership, the new CEO Su Lifeng also announced that he would enter the semiconductor industry, trying to compete with Nvidia in the AI boom.
In addition to the Silicon Valley giants, the new U.S. export restrictions to China have also cost Nvidia heavily, losing 25% of its total profits in China.
As soon as this news came out, Nvidia's market value evaporated by $53.5 billion overnight.
The absence of Nvidia has also led to a gap in the Chinese market, which has undoubtedly benefited Huawei and other latecomers.
In order not to be "stuck" by the United States, the Chinese government decided to raise 300 billion yuan to set up a national fund to accelerate the development process of semiconductors.
However, as a veteran of AI chips, Nvidia's products can still easily crush all competitors in terms of maturity and cost performance.
Having experienced the wave of virtual currency, metaverse and AI, Huang's continued success as an AI leader still depends on his mastery of the latest technology and his judgment of industry trends.
Now in his prime of life, Huang has been at the helm of Yingwei for more than 30 years, and with a net worth of $23.8 billion, he has no intention of retiring at all, and still maintains full enthusiasm and energy to work every day, eager to lead LinkedIn Weida into the next 30 years.
Steve Jobs once said, "Don't let the noise of other people's opinions drown out your inner voice, the most important thing is to have the courage to chase your heart and follow your intuition." ”
If you feel like you're getting nothing out of nowhere, don't panic, Huang didn't start his business until he was 30 years old.
If you think you're working hard and hard, don't be a cow, Huang is still at the top of the wave at the age of 60.
-END-
Resources:
1 . "Jensen Huang: When you can't tell if you're a hunter or a prey, just run," Southern People Weekly
2. "Is Artificial Intelligence "Played"?" Nine Sides
3. "Silicon Valley Hard Currency Nvidia, Rose 500 Billion Overnight" South Wind Window
4. "Nvidia jumped over "trillion dollars", what did Jensen Huang do right?"
5. "60-year-old Jensen Huang: I hope to work for another thirty or forty years, and then become a "robot"" The Paper
Author: Maple Yeqing
Editor: Willow Leaf chatter