1 trillion dollars! The world's highest market value chip company has been born, close to two TSMC! Why did it win the big trend?

Edited by: Cheng Peng, Gai YuanyuanBefore the US stock market on May 30th,Nvidia's market value broke through trillions of dollars, becoming the first chip company in the history of the US stock market to reach a market value of $1 trillion, creating history.Image source: First FinancialIn the recent wave of artificial intelligence, NVIDIA is undoubtedly at the forefront of the trend, with its momentum even surpassing ChatGPT's owner Microsoft recently

Edited by: Cheng Peng, Gai Yuanyuan

Before the US stock market on May 30th,Nvidia's market value broke through trillions of dollars, becoming the first chip company in the history of the US stock market to reach a market value of $1 trillion, creating history.

Image source: First Financial

In the recent wave of artificial intelligence, NVIDIA is undoubtedly at the forefront of the trend, with its momentum even surpassing ChatGPT's owner Microsoft recently. One key reason why NVIDIA can stand out is its widely sought after chip products in the field of artificial intelligence,A100 chips and higher generation H100 chips, currently these high-end chips and corresponding graphics cards are difficult to obtain.

Zhang Yi, a senior researcher at Microsoft Asia Research, recently exclaimed in a podcast that the strange scene of the entire Earth not being able to produce enough A100 chips is now happening. A year ago, almost no one expected this situation.The A100 chip launched by Nvidia in 2020 is now available at a price but not in the market, and the H100, which has become popular on ChatGPT, has been frantically snapped up by large companies.This has also made Nvidia's performance soar all the way, and its stock price has continued to rise.

Brannin McBee, founder and CEO of CoreWeave, a startup in the field of artificial intelligence, couldn't help but sigh:H100 is one of the most scarce engineering resources on Earth.This statement is enough to give a glimpse of Nvidia's current prosperity.

NVIDIA Achieves the World's Most Valuable Chip CompanyClose to two TSMC

On the evening of May 30th, the US stock market opened, and NVIDIA's stock price rose 3.8% to $404.25 per share. As of press release, NVIDIA reported a price increase of 5.29% to $410.06, with its market value exceeding $1 trillion, reaching $1.014 trillion.

According to the companiesmarketcap website, Nvidia ranks 6th in the world with a market value of $1 trillion and is currently the chip company with the highest market value,Close to two TSMC5340180%

On the news, according to foreign media reports, recently investment bank JPMorgan Chase stated in its investment report that with hardware products such as GPUs and network products,Nvidia will occupy up to 60% of the artificial intelligence (AI) product market this year.

This trend is also reflected in Nvidia's first quarter financial report. On May 25th, Nvidia announced its first quarter results of this year, with revenue of 7.19 billion US dollars, a year-on-year decrease of 13%; The net profit was 2.04 billion US dollars, a year-on-year increase of 26%. But during the same period,Its data center business revenue reached a record high of 4.28 billion US dollars, a year-on-year increase of 14%, accounting for 60% of its total revenue;The revenue of the gaming business was $2.24 billion, a year-on-year decrease of 38%, accounting for 31% of its total revenue.

After releasing a significantly better than expected revenue forecast last week, Nvidia's stock price skyrocketed by nearly 30% overnight.

At the performance call, Nvidia told analysts that many cloud companies are competing to deploy AI chips. The demand for graphics processing unit (GPU) terminals for personal computers (PCs) was "stable" in the first quarter. Performance growth is coming from the data center business.The company has locked in significant growth in data center chips and plans to significantly increase supply in the second half of the year.

At the same time, the company provides guidance on the next quarter's performance,It is predicted that the sales revenue in the second quarter will reach 11 billion US dollars,Floating up and down by 2%,Far higher than the expected 7.2 billion US dollars.This also led to a 25% surge in NVIDIA's stock price in after hours trading, increasing its market value by nearly $200 billion.

According to Cailian News Agency, analysts point out that for Nvidia, the business opportunities brought by this AI boom will be far more important and sustainable than cryptocurrencies. Nvidia's chips and software can meet the computationally intensive needs of generative AI, and its product richness is unmatched in the industry. According to UBS analysts' estimates,Developing the chat robot ChatGPT requires approximately 10000 Nvidia GPUs.

The popularity of artificial intelligence this year has brought about a huge explosion in chip demand. As a one-stop solution provider, Nvidia provides training services that can help AI products train a large amount of text, images, and videosGPUIn the AI big model competition, he mastered the "lifeline" of computing power supply, thus becoming a major winner in this year's AI concept craze.

Currently, NVIDIA is the dominant player in the GPU market, with a global independent graphics card market share of up to 80%,Its high-end GPUs such as H100, A100, and V100 occupy the vast majority of the AI training market share. Due to the ability of GPUs to provide computing power for large language models, market demand for them is on the rise, and there have been reports of a shortage of Nvidia chip supply in the market before.

Introducing AI supercomputers:500x memory expansion!

The stock price rise of chip manufacturers this year is largely driven by the demand for computing power from artificial intelligence. Nvidia is standing at the forefront of "supercomputing power".

According to Pengpai News, on May 29th,NVIDIA CEO Huang RenxunIn the opening speech of the COMPUTEX conference held on Monday,Published the supercomputer DGXGH200,Huang Renxun called it "integrated with EvergrandeThe most advanced accelerated computing and network technology

Image source: Visual China

The DGXGH200 is a large memory AI supercomputer that can be used to support generative AI and data processing giant models. It is also the first supercomputer to pair the GraceHopper superchip with the NVLinkSwitchSystem, which enables all GPUs in its system to operate together as a whole.

It is reported that the DGXGH200 has a total of 256 GraceHopper super chips, which can provide 1exaflop performance and 144TB of shared memory. Compared with the previous generation DGXA100 launched in 2020,Memory has expanded by nearly 500 times.

In terms of partners, Google Cloud, Meta, and Microsoft are expected to be among the first to access DGXGH200 for generative AI products. Next, NVIDIA plans to use the DGXGH200 design as a blueprint for cloud computing companies and other large-scale enterprises.

On this basis, Nvidia also plans to build an AI supercomputer called "Nvidia Helios" based on the DGXGH200. This supercomputer will be equipped with four DGXGH200 systems, including 1024 GraceHopper superchips, and is expected to be launched by the end of this year.

According to First Financial, Nvidia is also building a large-scale generative AI supercomputer, Israel-1, which will be deployed in Nvidia's Israel data center.

Huang Renxun stated that people are now at a turning point in a new computing era, where accelerated computing and artificial intelligence have been accepted by almost all computing platforms and cloud service providers around the world.

AIAI

Why did Nvidia win the big trend?

Why is Nvidia the only player in the field of artificial intelligence with millions of chips in the world? And why can Nvidia, a company that has always dominated graphics cards, establish such a big presence in the fields of deep learning and artificial intelligence?

According to Cailian News Agency, in 1999, the emerging Nvidia first introduced the concept of GPU. Prior to this, CPU manufacturers, including Intel, firmly believed that graphics processing was the job of the CPU, and the more tasks the CPU did, the better. The idea of separating graphics work from another attached processor was extremely weak.

At that time, Japanese manufacturers engaged in gaming had the most say in the field of graphic applications. The CPU of Japanese hosts is very strong, and most development work is focused on the CPU, so GPUs do not have much market space.

The turning point is that Microsoft, who is not convinced, wants to challenge its industry-leading position as a Japanese manufacturer by developing DirectX, a standardized API graphics interface. Afterwards, a large number of graphics functions were ported from the CPU and transferred to the GPU. In addition, with the launch of another Microsoft product, Xbox, the combination of CPU and GPU has broken the dominance of CPU chips in the industry.

Nvidia was one of the few companies in the hardware industry that followed the Microsoft flag back then, and went all the way down the GPU path.

Afterwards, Microsoft pushed for another revolution by introducing unified rendering technology, which allowed GPUs to merge the vertex calculation and subsequent rendering steps of graphic drawing. It has partnered with another well-known company in the graphics card industry, ATI, to successfully apply GPUXenos technology.

Unified rendering is just a step in graphics applications, but it has brought a completely different development path to Nvidia, which can be said to be the starting point for Nvidia's later development in GPU and even its involvement in deep learning.

After seeing the unified rendering architecture, Nvidia decisively overturned its previous GPU architecture and started over. Its GPU stream processors have been carefully grouped into small stream processors that can run independently, solving the problem of stream processors being bound and forced to idle.

This laid the foundation for NVIDIA's revolutionary CUDA architecture.Due to NVIDIA's stream processors being very independent and standard units, they are easy to control and schedule, allowing tasks that could only be processed serially to be processed in parallel. This greatly reduces the difficulty of programming.

At the same time, NVIDIA's competitor ATI did not invest in hardware architecture changes in the early stages, as they continued to use past serial designs, resulting in higher sunk costs, making innovation increasingly difficult and expensive. In the end, NVIDIA successfully squeezed out of the graphics card market.

Afterwards,Nvidia also introduced the TensorCore computing unit concept in 2017, which is specifically designed for deep learning and supports lower precision operations, greatly saving model computing power.

This dedicated acceleration unit has largely squeezed out the space for CUDA to handle deep learning, but it has also caught Nvidia's competitors off guard, making AI specialized chips no longer attractive. As a result, NVIDIA GPU coincidentally became the most recognized hardware in the AI field.

In 2003, "Fast Iteration, Nvidia, which keeps trying and making mistakes, has made an unpopular project. It has developed a Soc chip that integrates the ARM based CPU with its own GPU. Since the Soc chip, Nvidia has released some chips every few years. In 2015, it launched TegraK1, which uses the Arm public CPU and its own Johannes Kepler based GPU. However, due to the unsatisfactory power consumption and heat, most users are tortured.

But industry insiders are very supportive of these setbacks.An investor once pointed out that while Nvidia is holding onto its GPU base, it is constantly extending its reach in new fields and allowing countless people who have bought its graphics cards to share costs with it. He also praised that although many things of NVIDIA, such as CUDA, could not be seen landing for a period of time, it established a complete ecosystem during the trial and error process and successfully stood out when a new wind struck.

This is also one of the reasons why NVIDIA GPUs defeated other chips and successfully took the AI dividend.On the one hand, GPUs have better versatility and are more adaptable to changes than specialized chips; On the other hand, NVIDIA has a complete ecosystem, making its GPU the most suitable choice at the moment.

In fact, when AI suddenly erupts, companies in the industry have no choice but to find that GPU is the best choice for simple and efficient running of generative AI models. A GPU originally used for playing games is unlikely to switch to running AI programs,Currently, only NVIDIA's GPU can run AI models.

edit|Cheng Peng Gai Yuanyuan

Proofreading|Duan Lian

Image source: Visual China

Daily Economic News Comprehensive Self Finance Associated Press, Pengpai News, First Financial, Public Information

Daily Economic News

Disclaimer: The content of this article is sourced from the internet. The copyright of the text, images, and other materials belongs to the original author. The platform reprints the materials for the purpose of conveying more information. The content of the article is for reference and learning only, and should not be used for commercial purposes. If it infringes on your legitimate rights and interests, please contact us promptly and we will handle it as soon as possible! We respect copyright and are committed to protecting it. Thank you for sharing.(Email:[email protected])