The deep reason behind Nvidia s skyrocketing 2 trillion

Mondo Digital Updated on 2024-02-26

Wen Wang Xinxi.

On the evening of February 22, investors all over the world were shocked by Nvidia's stock price. As of February 22 local time**, Nvidia reported 785$38, up 164%, and the market value increased by $273.3 billion overnight(about 2 trillion yuan)., which rose twice last year, has risen by 60% this year, and its market value has surpassed Amazon and Google, second only to Microsoft and Apple.

Nvidia's current market capitalization is as high as 19 trillion US dollars, this figure is about the same as the GDP of Italy and Canada, which are among the top ten in the world, and they are already rich.

Nvidia soared by 2 trillion yuan overnight, and not many people understand the reason behind it.

NVIDIA has always been in a key position in the development and demand of AI and is the promoter of the industry

Nvidia's ability to stand out in the chip industry, where there are many strong players and even the state as a combat unit, is inseparable from the unique vision of Nvidia CEO Jensen Huang.

In the early 90s of NVIDIA's establishment, Lao Huang aimed at the gap in the game market with a unique vision, and developed a product that is different from all graphics chips on the marketNV1, which integrates graphics cards, sound cards, handle drivers and other functions, which gave rise to the first graphics revolution and brought the birth of GPUs.

After becoming a major player in the chip industry, in 2004, he began to do software development platform, and the CUDA computing platform was born, which improved the parallel computing performance of the graphics card.

The efficiency of parallel computing has been greatly improved, laying the foundation for fields that were previously completely unimaginable, such as deep learning, image recognition, autonomous driving, and AI.

2012 was a crucial year, Nvidia GPU became a god, Alex Krizhevsky used GPU for deep learning, won the ImageNet competition after a few days of training, and he used $1000 hardware + CUDA platform to make a computation that even a million-dollar supercomputer could not complete at that time.

After that, giants such as Microsoft and Google frantically placed orders for Nvidia's GPUs, and by 2015, almost all companies involved in image processing and artificial intelligence were using Nvidia's products, and its chips were widely used in search, image recognition and artificial intelligence. Later, Lao Huang abandoned smartphone chips, bet on artificial intelligence and GPU, and became the only one in the world in high-end artificial intelligence chips.

Since then, NVIDIA has continued to introduce new products and technologies, such as RTX series GPUs, NVIDIA **ATAR Cloud Engine (ACE), etc., which have been used in gaming, professional visualization and autonomous driving, further broadening the company's market and revenue**.

In November 2022, ChatGPT launched 3After version 5, artificial intelligence exploded overnight, and a large number of companies are grabbing the best AI chips, which has made NVIDIA's data center business grow rapidly, especially the application of its GPU (graphics processing unit) in AI training and inference, which has brought it significant revenue growth. In general,With the development of AI and the surge in demand, NVIDIA has always been in a key position.

The first large-scale model SORA exploded, and Nvidia stepped on all the outlets, driving growth expectations

Nvidia reported fourth-quarter revenue of $22.1 billion — data center revenue of $18.4 billion and automotive revenue of $2.4 billion$8.1 billion, with $2.9 billion in gaming revenue, almost all of which exceeded expectations.

At present, the first large-scale model SORA exploded, further detonating the global demand for computing power, and NVIDIA occupies a global leading position in the artificial intelligence (AI) product market, with a fairly high premium in the high-end chip market, which directly detonated its market value.

Historically, Nvidia has stepped on almost all the outlets, around 2015, when VR was on fire, Huang Jenxun released the 1060, a product with soaring performance, and before that, the industry simply could not support the performance needs of VR.

In 2016, autonomous driving became popular, Huang Jenxun customized an autonomous driving computing card for Tesla, and in 2018, the blockchain became popular, and Lao Huang transfused blood to the mine boss overnight, and the 3000 graphics card was dried to tens of thousands.

In 20 years, the metaverse was on fire, and in 22 years, it was ChatGPT, and the demand for Nvidia's high-performance GPUs has soared, and now it is the first big model, adding fire to AI, and Nvidia has finally soared.

At present, around the world, both large technology companies and leading institutions are actively investing in AI technology to improve data processing capabilities and develop intelligent applications.

There has been a surge in demand for high-performance computing across the globe, which has led to a massive demand for Nvidia's high-performance GPUsIn the AI computing power war, NVIDIA's GPU has become the key hardware support.

Nvidia's madness not only drove the A-share AI concept chicken dog**, but also drove Japanese chip stocks to soar, and European technology stocks also soared.

Nvidia still has potential in the Chinese market: the large-scale model gold rush, Nvidia is a shovel seller

In short, behind the crazy rise of NVIDIA, it is due to the fact that today's era has entered the era of computing power, computing power is the "base" of artificial intelligence development, the bottom layer of the large model is AI (computing power) chips, and ChatGPT is the result of relying on a large number of NVIDIA GPUs for intelligent computing.

Following the explosive growth of computing power demand brought by ChatGPT, SORA further detonated the global demand for computing power.

Now that the major tech giants are developing large models, accelerated computing and generative AI have reached a tipping point, in the words of Nvidia CEO Jensen Huang.

Even looking back at the past history, the breakthrough of artificial intelligence is the result of the efforts behind NVIDIA.

The market value of Open AI has increased by 6 times in less than a year, from 30 billion to 90 billion now. But in the context of this madness, the only region of Nvidia is China. Limited by U.S. chip sanctions, only castrated versions of Nvidia chips can be used in China.

And if the Chinese market is not restricted one day, can you imagine how much Nvidia's revenue will grow?

At the moment when ChatGPT is in full swing in China, a large number of domestic computing power models have appeared in our market in the past year. But the bottom layer behind this has left the ability of AI chips, and in the field of AI chips, NVIDIA is a monopoly and a leading state.

Nvidia has no benchmarks in the world, and neither does it in China

Globally, there is no company that Nvidia really has to compete with in terms of strength and business.

In the Chinese market, in the technology world, many technology companies in the United States, from Facebook, Amazon, Google to Apple, etc., have benchmark companies in China. But Nvidia didn't. According to Reuters, Huawei's Ascend series of chips is a competitor to Nvidia's AI chips. Huawei's Ascend 910B chip, launched last year, is seen as a Chinese alternative to Nvidia's A100 chip, which was launched three years ago.

In the field of AI chips and computing power, Huawei competes with Nvidia in terms of first-class artificial intelligence chips, but there is still a distance between it and NVIDIA in terms of product layout, the company's business direction and the overall strength of AI chips, and it is not a benchmark company at present.

Another company that will be talked about in the industry is Moore Threads, but in fact, the problem with Moore Threads is that the independent technology is not enough, and the MUSA architecture of Moore Threads is actually the IP architecture of the mobile phone GPU authorized by the British Imagination company. It has been established for just 3 years and has quickly launched a series of products such as S10, S30, S50, S60, S80, S3000 and so on, but it mainly relies on the small repair and modification of the British mobile GPU core, focusing on the stacking of core stacking indicators, and the performance and efficiency of the product are obviously lagging behind.

According to NVIDIA's leading trend, it is worth thinking about whether the gap between China and the United States in the field of AI will widen. This is why yesterday, the state issued a document requiring central enterprises to put artificial intelligence in an overall work for overall planning.

For example, in the past, Tesla, as the leader of electric vehicles, experienced a skyrocketing market value, and Musk also became the richest man in the world, but with the rapid growth of BYD, Tesla's imagination and market value growth space were suppressed.

At present, there is no series of platforms similar to CUDA and NVIDIA hardware deep binding in China, AI chip hardware technology and software technology are very far behind, many domestic GPU manufacturers are still adopting the strategy of compatibility with CUDA open source framework, in short, the current domestic manufacturers are still in the initial stage, there is no control autonomy at the bottom of the software and hardware, and there is no target company in China or even the world, which means that in the global market, in a short period of time, the stronger the stronger the trend is becoming more and more obvious, The gap may even widen, which is why the capital market is optimistic about Nvidia's future performance growth.

Judging from this trend, behind the soaring price of NVIDIA, we need to see a sense of crisis, and the investment and research and development of domestic computing chips should be stepped up.

Author: Wang Xinxi, TMT Senior Commentator This article is rejected without permission**.

Related Pages