This three-person game, with Zifeng Su, Jen-Hsun Huang and Zhongmou Zhang, also epitomizes the most current characteristics of the AI industry.

On June 14, the global AI community was waiting for a launch event. At the launch event, AMD's Chairman of the Board and CEO, Zifeng Su, stood in the spotlight wearing a blue stand-up jacket and a dry, short haircut, and intensely unveiled a number of new AI hardware and software products, including the new MI300X, a GPU (graphics processor) designed for large language models, taking the initiative to challenge the king of this field, Nvidia.The MI300X speeds up generative AI processing with up to 192GB of memory, more than the 120GB of memory in the Nvidia H100 chip, meaning it can train large language models with larger parameter sizes compared to the Nvidia H100 chip.
　　When NVIDIA won the lion's share of the AI computing market with its GPUs, there was always keen discussion about when AMD would launch a competitive product. Now, Suzy is here, and taking the world by storm.
　　The storm has involved the world's three largest chip companies: AMD has taken the initiative to invite Nvidia to battle it, Nvidia's stock soared that night, and TSMC, which is AMD's manufacturer and "behind the scenes", has received another large order.
　　The three companies at the helm, Zifeng Su, Jen-Hsun Huang and Zhongmou Zhang, are three of the most high-profile Chinese-American entrepreneurs in the global AI chip community. This three-man board also epitomises the AI industry at its most epochal.
　　NVIDIA's GPUs are seen as the best products for training big AI models, holding over 60% of the market share; AMD is considered by US investment banks to be NVIDIA's strongest competitor; and TSMC is the manufacturer of many chip designers such as NVIDIA and AMD, whose annual revenue accounts for 30% of the global semiconductor output.
　　From November 2022 to the present, the worldwide start-up frenzy of big models has been triggered with the introduction of ChatGPT. And behind the big-model frenzy, the tug-of-war for AI computing power is pushing these three companies to the centre of the tech stage.
　　Half a month ago, Nvidia broke $1 trillion in market capitalisation, becoming the world's first chip company to exceed $1 trillion in market capitalisation; not long ago, Warren Buffett made a comment on TSMC at the Berkshire. Not long ago, Warren Buffett at the Berkshire Hathaway annual meeting praised TSMC, saying that no company in the chip industry could match it.
　　The global chip industry mainly follows the following three development models: from chip design to manufacturing, such as Intel and Samsung; only chip design and R&D, manufacturing by the foundry, such as AMD, Qualcomm, Nvidia, etc.; focus on chip design companies to complete the foundry manufacturing, do not do their own design, such as TSMC, SMIC (52.270, -2.79, -5.07%), etc. .
　　Zifeng Su, Jen-Hsun Huang and Zhang Zhongmou have come from different places and together they are moving to the centre of the world AI chip battlefield.
　　Zifeng Su: Leading AMD back from the dead
　　Zifeng Su has the backbone to take the fight to Jen-Hsun Huang. Over the past decade, she has personally rescued AMD, which was once on the brink of bankruptcy, and has seen its share price increase nearly 30 times in less than a decade.
　　A company that predates Silicon Valley, AMD started making microprocessor chips for IBM in the early 1980s and at one point outsold Intel's in-house processors to take about a quarter of the server chip market. However, the good times did not last long, and when Suzi Fung took over in 2014, AMD was saddled with $2.2 billion in debt and even had to sell its own campus to rent it instead.
　　Born in 1969, the year AMD was founded, Zifeng Su was born in Tainan, Taiwan, and emigrated with her father to the U.S. at the age of three. In 1986, at the age of 17, Zifeng Su enrolled at the Massachusetts Institute of Technology to major in electrical engineering. When people asked her why she chose this major, she said, "Because I heard it was the hardest major." She began studying silicon technology in her sophomore year and spent her time at various factories to learn and practice the wafer manufacturing process. At MIT, Zifeng worked her way up to a Ph.
　　After graduating with his PhD in 1994, he joined Texas Instruments Semiconductor Process and Components Center as a technical specialist. A year later, she joined IBM's R&D department, where she was responsible for the development of copper wafer processes, and later served as head of IBM's R&D department and special assistant to the CEO, working for IBM for 13 years.
　　During her time at IBM, she was a researcher who helped design a semiconductor chip that used copper circuits rather than traditional aluminium circuits, resulting in a 20% increase in chip speed, and IBM executives quickly spotted her talent. to work as a technical assistant.
　　In January 2012, Zifeng Su joined AMD as Chief Operating Officer, Senior Vice President and General Manager of Global Business, etc. In June 2014, during AMD's reorganisation, Zifeng Su became CEO, in charge of AMD's large marketing business, and also became the first female CEO in AMD's history.
　　She became the first female CEO in AMD's history, and was affectionately known as "Mama Su" by AMD engineers. The day after she became CEO, she gave a speech to frustrated AMD employees on an all-hands conference call. She said, "I believe we can build the best products."
　　It was the first step in Suzi Fung's overhaul of AMD, and she had three things in mind that she had to do: create great products, deepen customer trust and simplify the company. "Doing just those three things is about keeping it simple," she said, "because if it's five things or 10 things, it's going to be hard."
　　AMD's share of the server market continued to decline, but Soo decided to focus most of her efforts on designing chip products anyway. She decided to prioritise the development of a new chip architecture called Zen.
　　This decision was successful when Zen was finally launched in 2017. By the time the third generation of Zen was released in 2020, it had become the market leader in terms of speed. the Zen architecture is now the basis for all of AMD's processors. Even when AMD ran out of chips to sell, Suzy spent years explaining to customers what she was doing.
　　In 2019, when Intel was struggling due to production delays and the loss of Apple as a customer, So jumped at the opportunity and led AMD to pick up customers such as Lenovo, Sony, Google and Amazon, which also allowed AMD to return to the top.
　　In 2021, Su became the first woman to receive the IEEE's highest semiconductor award, the Robert N. Noyce Award, and in 2022, she led AMD's $48.8 billion acquisition of the chip company Ceres, which produces programmable processors that help speed up tasks such as video compression.
　　As part of the deal, Ceres CEO Michael Peng became AMD's president and head of AI strategy. This also gives Sozi Fung more leverage to mount a challenge to Jen-Hsun Huang.
　　In June 2020, an Associated Press survey showed that Sozifeng was paid $58.5 million (about RMB 419 million) a year, with a base salary of $1 million, a performance bonus of $1.2 million and stock worth $56 million. So became the first woman among the world's most profitable CEOs and an indispensable key player in the chip market.
　　Jen-Hsun Huang: Entering the trillion dollar club
　　The year Su left for the US, nine-year-old Jen-Hsun Huang was also sent from Taipei to the US by her family to attend a rural boarding school in Kentucky.
　　This was a dangerous lesson in young Huang's life. He later recalled, "The school was really more like a juvenile reformatory, where every child had a knife and classmates were even tattooed all over their bodies." Instead, however, this taught Huang Renxun to be strong and adapt to his environment.
　　Two years later, Huang Renxun enrolled in a regular school. In addition to excelling in his studies, he also took third place in the nation in doubles for table tennis. But he was more technically inclined, and at the age of 16 he enrolled at the University of Oregon, where he studied electrical engineering and aspired to become a global graphics emperor while still in college.
　　After his undergraduate studies, Jen-Hsun Huang obtained a master's degree at Stanford University. In 1993, at the age of 30, he co-founded Nvidia with two engineers.
　　Unlike the chips made by Intel and AMD, Jen-Hsun Huang focused more on graphics chips that allowed games and graphics to run smoothly, as personal computers were just coming into the home at that time, but entertainment was still missing and could not meet the needs of running games. However, no one in the outside world was optimistic about GPUs as a new direction at that time.
　　Prior to 1999, Jen-Hsun Huang had launched two chips. At the end of 2000, Nvidia took its old rival 3dfx under its wing for US$110 million. In the same year, Jen-Hsun Huang challenged chip giant Intel.

Prior to 1999, Jen-Hsun Huang had launched two chips. But by betting on the wrong technology, Jen-Hsun Huang depleted the company's early investments and the company shrank from over 100 to 30 people. at the end of 2000, Nvidia took old rival 3dfx under its wing for US$110 million. In the same year, Jen-Hsun Huang challenged chip giant Intel.
　　In contrast to Intel founder Gordon Moore's "Moore's Law", Jen-Hsun Huang proposed "Huang's Law", that is, Nvidia's products are upgraded every six months, doubling the functions, such a technology update speed than "Moore's Law "This is two times faster than Moore's Law.
　　In late 2006, Jen-Hsun Huang opened up the GPU to software developers with the introduction of the CUDA platform, which allowed developers to apply the computing power provided by NVIDIA for purposes other than graphics. Although the results were minimal at first, as the era of artificial intelligence dawned, developers soon realised that GPUs were excellent at supporting the complex computations of modern AI systems.
　　Later, NVIDIA GPUs and the CUDA programming language became the foundation and standard for AI development training. This came as a surprise to even Jen-Hsun Huang. He mentioned in an exclusive interview with Forbes in 2016 that he had expected GPUs to be used in areas other than gaming, but never thought they would move to deep learning applications.
　　Today, Nvidia's products are the hard currency in global AI computing power. Jen-Hsun Huang had personally handed over the world's first AI supercomputer, DGX, to OpenAI, which developed ChatGPT. since then, the global competition of AI companies' computing power has also become a competition of how many NVIDIA GPUs there are.
　　In the past three years, Nvidia's market value has soared from US$150 billion to US$1 trillion, and Jen-Hsun Huang has been dubbed the "Godfather of AI". According to market research firm Tiburon Consulting, 60% to 70% of the more than 1.2 million AI servers shipped this year are equipped with Nvidia's GPUs.
　　Zhang Zhongmou: Making 30% of the world's computing power
　　In 1997, when Jen-Hsun Huang was still struggling to find the life of his business, another entrepreneur from Taiwan, China, reached out to him.
　　At that time, Jen-Hsun Huang had just developed his third generation product, RIVA128, which was approved by Microsoft. But the next challenge in front of him was how to quickly mass produce the product. At the time, it would cost US$100 million to build a fab, which was astronomical for a start-up company.
　　TSMC founder Chang Chung-Mou agreed to help Jen-Hsun Huang with OEM production.
　　When Jen-Hsun Huang was one year old, Chang Chung-Mou had already obtained his PhD in Electrical Engineering from Stanford University. Interestingly, after graduation, he intended to work for Ford Motors, but chose to work for a chip company because the monthly salary offered to him by Ford was US$1 less than a chip company, and he subsequently joined Texas Instruments, where he was promoted to general manager of the integrated circuit division the following year, and has worked here for more than 20 years since then. in 1985, after losing out as CEO of Texas Instruments, the 54-year-old Chang Chung-Mou returned to Taiwan, China, from the United States to found TSMC .
　　While still working at Texas Instruments, Chang had a radical idea: when demand for chips rose, chip design and manufacturing should be separated because the companies designing chips lacked the expertise to produce semiconductors. As technology advanced and transistors shrank, the cost of manufacturing equipment and R&D would rise, and only companies that produced large numbers of chips would have a cost advantage.
　　However, at the time, Texas Instruments, Intel and Motorola were all developing and producing their own. AMD founder Jerry Sanders even shouted the classic quote, "A real man has to have a fab". There was no support for Zhang Zhongmou's idea, and more crucially, part of the chip companies would fear that their design ideas and creativity would be copied.
　　But in the end, it turned out that real men can indeed do without fabs.
　　To allay the concerns of chip design companies, Chang promised that "TSMC will never design chips, it will only manufacture them." Coupled with the chip design revolution at the time, which made it easy to separate design and manufacturing, Chang caught the perfect opportunity, which the industry even likened to the chip industry's "Gutenberg moment".
　　Today, the new wave of chip entrepreneurs in Silicon Valley do not have fabs.
　　The chips of companies such as Nvidia, AMD and Apple are mainly made by TSMC. It is also because of the choice of Nvidia and a host of other chip design companies that TSMC has achieved its glory. in 2017, TSMC surpassed Intel in terms of market value and became the world's number one chip manufacturer. It was also the year that Chang Chung-Mou announced that he would retire from any position at TSMC the following year.
　　TSMC's Business Report to Shareholders shows that TSMC's revenue will account for 30% of global semiconductor (excluding memory chips) output in 2022. In terms of shipments, in 2022, TSMC offered 288 different process technologies and produced 12,698 different products for 532 customers.
　　After Zhang Zhongmou's glorious retirement, TSMC rode on the AI east wind to leave its competitors far behind. His successor is increasingly aware that as the industry begins to move into the AI era, a smarter and more connected world will create a strong demand for computing power and low-power computing, and they will be greeted by an even larger market.