A few days ago, Nvidia's market capitalization exceeded $4.2 trillion, once again becoming the world's most valuable company. In less than four months, Nvidia's stock price has skyrocketed by 83%; The 14-day Relative Strength Index (RSI) broke above 80 at one point last Thursday.

So, what is causing Nvidia's stock price to rise so wildly?

According to CCTV News, Nvidia CEO Jensen Huang announced two important developments in an interview with Chinese media:

The United States approves H20 chips to be sold to China: This means that Nvidia can continue to sell its high-end chips to the Chinese market, which is a major positive for the company.

RTX coming soon New Pro graphics card: This new graphics card will further enhance Nvidia's competitiveness in the high-performance computing and gaming markets.

This series of moves has once again pushed the close relationship between NVIDIA and China's large models and AI industry to the forefront, triggering investors to think about the relationship between the two.

Weida: The "double-edged sword" of China's AI

First, let's take a look at Nvidia's contribution to China's AI field. There is no doubt that NVIDIA is a well-deserved leader in the global AI field, and its GPU chips provide a strong computing power foundation for AI computing and are indispensable key hardware for large model training.

In China, many mainstream large model companies and AI companies are highly dependent on Nvidia's GPUs. For example, companies such as ByteDance and Tencent will use Nvidia's H100, H20 and other high-end chips when training large-scale language models; These chips not only accelerate the model training process, but also help enterprises quickly develop various AI applications, which are widely used in natural language processing, computer vision and other fields.

In addition, through cooperation with Chinese companies, NVIDIA has promoted the application of AI technology in practical scenarios such as smart factories and logistics, providing an important boost to industrial development.

However, there are huge risks behind it, and despite the many benefits that Nvidia has brought, there are also many risks hidden behind this close cooperation:

1. Serious technology dependence

Due to Nvidia's monopoly in the GPU field, it is difficult for Chinese companies to find alternatives in the short term. If there is a supply chain problem, such as the US government tightening export controls for political purposes, the business development of the company will be hit hard.

2. A large outflow of profits

Another problem is that Chinese companies need to pay high fees for purchasing Nvidia chips, which leads to a large amount of profits flowing to Nvidia, compressing the profit margins of local companies. In this way, the funds used by enterprises for independent research and development and long-term development are reduced.

3. Innovation is limited

Over-reliance on Nvidia's technology may lead to a lack of motivation for Chinese companies to innovate the underlying technology. Enterprises may fall into the cycle of "introduction, use, and re-introduction", and it is difficult to form an independent and controllable core technology system, and they are always in a passive position in global competition.

| Dual impact on Chinese companies

1. Restrict development autonomy

First of all, from the perspective of development autonomy, Nvidia's existence does form obvious constraints on China's large models and AI companies.

Due to Nvidia's monopoly in the GPU field, Chinese companies have to rely on their products for large model training and AI computing. This leads to:

R&D progress is subject to others: The R&D progress and technology update rhythm of enterprises depend entirely on NVIDIA's product supply and technology upgrades.

Increased policy risks: When the U.S. government introduces policies that are not conducive to Chinese companies' access to advanced chips (such as banning the sale of H20 chips), companies' large model training and upgrade plans may be forced to be shelved or delayed.

In addition, due to Nvidia's dominance in the global market, Chinese companies often lack a say in cooperating with it and find it difficult to obtain more favorable terms for cooperation. This further limits their room for development.

2. Impact the local market

Next, let's see how Nvidia hits the local market. With its technology and brand advantages, Nvidia has attracted a lot of resources and attention in the Chinese market. Many investors are more inclined to invest in companies that use NVIDIA technology because they look more "successful". This has brought greater difficulties to local innovative enterprises in terms of financing.

3. Market share squeeze

In the application market, Nvidia's products and technologies also occupy a large share, squeezing the market space of local companies. Especially in some high-end AI computing fields, it is difficult for local companies to compete with them.

For example, companies such as ByteDance and Tencent almost always choose high-end chips such as Nvidia's H100 and H20 when training large-scale language models, but once there is a problem in the supply chain, these companies will face huge challenges. In April this year, the United States banned Nvidia from selling H20 chips to the Chinese market; Although permits were reinstated in July, this has highlighted the instability of supply and created great uncertainty for businesses.

Technological innovation is not to take a flat road, but to climb a slope

In the face of a "global hegemon" like Nvidia, China's AI chip companies can be said to be struggling on a road full of thorns.

Taking Huawei's Ascend series of chips as an example, a lot of progress has been made in recent years. For example, Ascend 910B and Ascend 920 are close to some of NVIDIA's products in terms of performance; In some specific scenarios, such as image recognition, inference computing, etc., the performance is even better.

But then again, technology and products are only the first step, and the real challenge lies in the ecology. Nvidia has built a very mature software ecosystem with decades of technology accumulation, such as the CUDA platform, which is like a "playground for AI developers", attracting millions of developers around the world. The more complete this ecosystem, the more willing developers are to use NVIDIA chips; The more people use it, the more difficult it is to replace; This creates a "high wall and deep pool", which is difficult for others to enter.

In addition to technical problems, the pressure of market competition is not small. Nvidia is not only strong in technology and ecology, but also in brand. When many customers choose a chip, their first reaction is to "choose a safe one", so Nvidia has reached the default option. There is still a gap between domestic enterprises in brand awareness and market recognition.

At the same time, the competition between domestic AI chip companies is also quite fierce. Everyone is competing for customers and orders in a limited market space, and the pressure is not small.

Recently, Nvidia CEO Jensen Huang visited China, once again bringing Nvidia's relationship with China's AI industry closer. This cooperation does bring some short-term benefits, such as the re-approval of H20 chips for sale.

But enterprises should also be soberly aware that cooperation can be there, but dependence cannot exist.

Of course, for investors:

pay attention to those enterprises with long-term layout in chip architecture, software ecology, algorithm collaboration, etc.; Pay attention to the development trends of domestic chip companies such as Huawei Ascend, Cambrian, Biren Technology, and Horizon.

In the next few years, whoever can build their own ecosystem may become the "game-breaker" of China's AI chips.