Two of the most powerful Chinese in the semiconductor world cast a shadow over Kissinger's path of counterattack

The shuffle of the Semiconductor World Power List is really happening. Intel has been sitting on the throne of the world's most expensive semiconductor company for many years, but in recent years, it has been surpassed by two semiconductor companies managed by Chinese people.

AMD surpassed Intel for the first time in mid-February due to its successful acquisition of Xilinx. The first time Nvidia's market value surpassed Intel's, it occurred on July 2, 2020, and became a major event that shocked the industry.

AMD and NVIDIA, two current high-profile semiconductor companies, are led by Su Zifeng and Huang Jenxun respectively, and the Chinese identity is the most distinctive commonality between the two CEOs.

Unlike the Gathering of Indian executives from Internet companies, Silicon Valley semiconductor companies other than Intel are mostly led by Chinese, so it is not unusual for Intel's market value to be overtaken by Chinese-led companies. However, the high prosperity of AMD and Nvidia has undoubtedly made Su Zifeng and Huang Jenxun the two most powerful Chinese in the semiconductor world.

At the same time, both companies are opening their bows to the former overlord Intel - the most advantageous desktop PC market is gradually cannibalized by AMD, the highly deterministic data center market has also been seized by NVIDIA, Kissinger with LinkedIn The Ter Jedi counterattack on the road, must fight Su Huang.

In addition to market competition, Su Huang's praise for his past achievements has made Kissinger envious. One brought AMD back to life, the other founded Nvidia and rewritten the semiconductor market with artificial intelligence, and was also named "Entrepreneur of the Year" by Fortune magazine at the end of 2020. Not only that, one is called Su Ma, the other is called Lao Huang, and obviously each has a good fan base.

Su Mama walked over the glass cliff

Su Zifeng joined AMD in 2011 and officially took over as CEO at the end of 2014. She holds a Ph.D. in Electrical Engineering from the Massachusetts Institute of Technology in the United States and has over 15 years of experience in the semiconductor industry prior to joining AMD.

Studies by management scholars in Europe and the United States have found that when women finally have the opportunity to prove themselves at the top, they often take over companies that are in a bad state. In other words, Su Zifeng seems to have broken the glass ceiling of gender discrimination in the workplace, but in fact stands in front of the glass cliff with the risk of falling.

What lies in front of Su Zifeng's eyes is the glass cliff. At that time, AMD was in a deep crisis, its market share was cut, its revenue fell by 40%, its stock price fell to about $3, and it had changed three CEOs in 5 years.

Su Zifeng said on his personal LinkedIn page: My passion is to lead the global technical team to bring outstanding products and solutions to market. What then helped Su Zifeng lead AMD to achieve a surprising reversal was precisely a series of blockbuster products that he led.

In 2017, AMD officially launched a high-performance CPU processor based on the new "Zen" architecture, and AMD released the EPYC series of high-performance data center processors. Under the leadership of Su Zifeng, AMD regained market share step by step and got rid of the brink of bankruptcy.

Many people in the industry also attribute Su Zifeng's success to its strategy of decisively transferring to A single TSMC. When competitor Intel was stuck in a process delay, Su Zifeng successfully persuaded shareholders to transfer orders from GF to TSMC, thus launching a more competitive processor and becoming the biggest beneficiary of TSMC's advanced process overtaking.

Su Zifeng has been trying to push the company's PC processor business to the peak, and the latest release data can support her another stage of victory. In intel's x86 CPU market, which once accounted for 90% of the market, AMD won 25.6% of the market in the fourth quarter of 2021, a new 15-year high.

Lao Huang continued to write Moore's Law with AI

The semiconductor industry has a high opinion of Huang Jenxun.

Corresponding to Moore's Law, which describes the speed at which chip performance improves, Huang's law, which predicts that GPUs will drive AI performance to double year by year, is named after him. TSMC founder Zhongmou Zhang also praised him as an extraordinary leader and said the world can expect his great achievements in the next decade.

Underpinning this accolade is the development of AI semiconductor products led by Huang Jenxun and his perfect grasp of growthable markets.

Jen-Hsun Huang founded NVIDIA in 1993, the company started with GPUs and focused on the gaming market. Later, GPUs were found to be well suited to the artificial intelligence market that pursued large-scale rapid processing, and Huang Reacted Quickly and immediately began to improve the supporting software.

Since around 2015, NVIDIA has been applying technology to artificial intelligence. The second quarter of 2020 was a turning point, with NVIDIA's data center revenue, mainly semiconductors for AI computing, surpassing the main game business for the first time.

Data center revenue is still growing steadily. In the fourth quarter of fiscal 2022, NVIDIA data center revenue reached $3.26 billion, an increase of 71% year-on-year. If Jen-Hsun Huang's prediction that the data center will become the basic unit of computing falls into reality, the potential of this market is immeasurable.

Heterogeneous computing acceleration

The semiconductor industry ushered in a big year of mergers and acquisitions in 2020, and it was NVIDIA and AMD that played the heaviest role. Nvidia plans to acquire Arm from SoftBank Group, which is expected to cost up to $40 billion. It is followed by AMD, which plans to spend $35 billion to buy Xilinx.

Whenever there is a wave of mergers and acquisitions, it often reflects changes in industrial technology and environment.

The dual background of the 2020 epidemic and geopolitical tensions has helped companies cope with uncertainty with mergers and acquisitions.

From the perspective of the semiconductor industry, after the market focus shifts to a new generation of information technology such as artificial intelligence, 5G and the Internet of Things, there are higher requirements for computing power. Processors of a single type and architecture are difficult to deal with, and the heterogeneous computing routes that combine different processor solutions have gradually become the consensus of the industry.

Both mergers and acquisitions ushered in very different fates last month, AMD bought the number one FPGA manufacturer as promised, and NVIDIA's acquisition of Arm was terminated under multiple resistances. Although the two mergers and acquisitions are headed for diametrically opposed fates, they can both play a role in promoting the transformation of heterogeneous computing and make the competitive situation around the Big Three clearer

Ultramicro Semiconductor, which already has CPU and GPU products for both personal computers and servers, combined with Xilinx's FPGAs, will naturally help it hit the data center and communications base station markets.

After huang Jenxun announced the failure of the merger case, he said in response to NVIDIA's heterogeneous computing that the future will span the three-chip strategy of CPU, GPU and DPU.

Huang Renxun said that NVIDIA still intends to continue to cooperate with Arm to accelerate the development of high-end CPUs. However, cooperation with Arm and the purchase of Arm are naturally quite different. Otherwise Nvidia wouldn't have tossed around for more than a year and ended up paying $1.36 billion in lost costs.

Also around heterogeneous computing, Intel has long been clear about the XPU strategy that includes CPUs, GPUs, IPUs, FPGAs, etc. Intel's recent intensive moves, with the launch of discrete graphics, co-development of ASIC-based IPUs with Google Cloud, and early purchase of the second largest FPGA vendor can be seen as efforts to build heterogeneous computing at the hardware level.

epilogue

Su Zifeng and Huang Jenxun took advantage of the victory to pursue, the other suffered setbacks, and finally met Kissinger in key markets such as data centers. Competition in the data center market is becoming fierce, which also provides a moment for these star entrepreneurs to show their charm.

In contrast, also from a technical background and in danger, Kissinger can learn from Su Zifeng how to fight back.

Intel, determined to restructure the wafer foundry business, has bought a tower on the one hand, and has also booked TSMC's 5nm/3nm production capacity in advance. Kissinger, who is no longer stuck in self-manufacturing and has come up with a more flexible IDM2.0 strategy, has the wisdom of Su Zifeng to make a decision.