Trillion Nvidia, evaporated 400 billion overnight
Chinese internet companies face the risk of cancelling their $4 billion NVIDIA high-end chip orders.Article | Chinese Entrepreneur Reporter Deng ShuanglinEditor | Li WeiHead Image Source | Visual ChinaOn the evening of October 17, the US Department of Commerce released an updated version of the chip export control provisions in Chinese Mainland, detailing and upgrading the 2022 rules
Chinese internet companies face the risk of cancelling their $4 billion NVIDIA high-end chip orders.
Article | Chinese Entrepreneur Reporter Deng Shuanglin
Editor | Li Wei
Head Image Source | Visual China
On the evening of October 17, the US Department of Commerce released an updated version of the chip export control provisions in Chinese Mainland, detailing and upgrading the 2022 rules. This version further tightened export controls on chips to China, andIt means that the high-performance chips of Nvidia, AMD, Intel and other American chip giants will be banned from selling to the Chinese Mainland market.
On October 7 last year, the United States issued export control measures against semiconductors and others, and put forward a number of detailed provisions, including restrictions on Chinese Mainland enterprises' access to high-performance chips, and defined the performance indicators of controlled hardware, known as "the most stringent rules in history". At that time, due to export control, Nvidia's A100, H100 and other GPU chips could not be exported to Chinese Mainland. In order to comply, Nvidia restricted the transmission rate and launched special versions A800 and H800 for the Chinese Mainland market; Intel has also launched Gaudi2 specifically for the Chinese market.
With the release of the ban,NVIDIA's stock price, with a market value of over a trillion US dollars, plummeted by nearly 5% on the same day,The market value evaporated overnight to over 53.5 billion US dollars (approximately 400 billion RMB).
Affected by the new ban, the civilian consumer market also quickly took action - on the afternoon of the 18th, Nvidia's top flagship graphics card RTX4090 began to be fully taken off the shelves. According to a search by "Chinese Entrepreneur" on e-commerce platforms such as JD and Taobao, official channels are currently out of stock, and third-party stores have a small amount of inventory, but the price has jumped from the original price of 12999 yuan to around 30000 yuan.
Source: Visual China
In response, NVIDIA stated, "We comply with all applicable regulations and strive to provide thousands of application products that support different industries. Given the global demand for our products, we do not anticipate that the new regulations will have a substantial impact on our financial performance in the short term
The new regulations may affect the company's development and delivery capabilities,This includes completing the development of certain products on time, providing support to existing customers of these products, or supplying these products to customers outside the affected areas
NVIDIA or Lost China?
Specifically, the United States has changed the parameter specifications to "Total Processing Performance" or "Performance Density", and chips above a certain value are subject to export control. The affected products include graphics processing units (GPUs), tensor processing units (TPUs), neural processors, memory processors, etc.
The additional licensing requirements for chip performance thresholds in the above-mentioned export control revisions will affect products such as Nvidia A800 and H800, and even cover the flagship product RTX4090, which is mainly used in gaming PC (personal computer) consumer products. Computer system products such as DGX and HGX that integrate the aforementioned chips are also covered by the new regulations.
Moreover, the new regulatory red line will have a certain impact on the mass production progress of domestic GPUs with AI computing power chip design, TSMC 7nm/5nm OEM, and AI computing power chip manufacturers, includingEnterprises such as Bi Ren Technology, Moore Thread, Cambrian, Suiyuan Technology, etc.
However, in response to the consumer market, the US has introduced an exemption allowing the export of chips for consumer applications such as laptops, smartphones, and gaming applications. For a small number of high-end game chips, license notification requirements are still in place, with the aim of improving shipment visibility rather than completely banning sales.
Nvidia mentioned that this may affect the company's ability to complete product development on time, support existing customers of restricted products, and supply restricted products to customers outside of restricted areas. In addition, the new regulations may also require companies to relocate some of their operations from the affected areas. In response, Nvidia stated that if customers need to purchase products affected by new US government regulations, the company may apply for a license for them, but cannot guarantee that the US government will approve or process the request in a timely manner.
Previously, some media reported that,Multiple Chinese internet companies have placed a total of $1 billion in orders with Nvidia in 2023,We are vying to deliver approximately 100000 A800s specifically for the Chinese market this year, and have booked $4 billion worth of A800s for delivery next year, including companies such as Tencent and Baidu.
Due to the new ban coming into effect in 30 days, Nvidia can still maximize the delivery of orders for this year before the end of the 30 day buffer period, butNext year's $4 billion order may face cancellation.
Zhang Xiaorong, President of the Deep Science and Technology Research Institute, pointed out to "Chinese Entrepreneur" that the impact of the ban on sales on NVIDIA is enormous, and banning the export of advanced AI chips to China will cause it to lose important market share and revenue sources. This move will have a negative impact on Nvidia's performance and profitability, and may lead to a continued decline in its stock price.
In May 2023, NVIDIA's market value exceeded trillions of dollars, making Huang Renxun the first Chinese entrepreneur in history to reach trillions of dollars. NVIDIA also became the first chip company in history to reach trillions of dollars. But it is not difficult to see from various data that,Most of the demand for artificial intelligence chips still comes from China,
Two months ago, NVIDIA submitted a beautiful second quarter financial report: revenue reached a historic high of $13.51 billion, a year-on-year increase of 101%; The revenue from the data center business was 10.3 billion US dollars, a year-on-year increase of 170%; The net profit reached 6.188 billion US dollars, a year-on-year increase of 843%.Among them, demand from China accounts for 20% to 25% of Nvidia's data center revenue.
However, NVIDIA issued a warning in a conference call after the financial report: given the strong global demand for NVIDIA products, if the United States imposes additional export restrictions on data center GPU chips sold to China, it is not expected to have a direct and substantial impact on NVIDIA's performance. However, in the long run, once such restrictions are implemented, This will permanently deprive American industry of the opportunity to compete and lead in one of the world's largest markets.
Five months ago, in a speech, Huang Renxun stated that chip export control measures could cause "significant damage" to the US technology industry. The export control order has made Nvidia's "hands tied behind their backs" unable to sell advanced chips in one of its largest markets. Huang Renxun emphasized that,China accounts for about one-third of the US technology industry market and cannot be replaced.
At present, it seems that with the implementation of more export control measures by the United States, it may create a window of opportunity for the rise of "Chinese chip", which may lead to greater competitive pressure on NVIDIA in the Chinese market.
The rise of domestic computing power?
In this year's global large-scale model computing arms race, Nvidia is undoubtedly the hottest arms dealer - computing power, algorithms, and data constitute the three elements of the AI era. Computing power is also an electric infrastructure, with how many GPUs and computing power it has become the key to external evaluation of the success or failure of large-scale model research and development and entrepreneurship. GPU is the core component for ChatGPT training and deployment, and Nvidia has almost no competitors in the GPU field.
Nvidia's monopoly on AI computing power has also left big model manufacturers struggling for a long time. According to industry evaluations, a ChatGPT requires 10000 NVIDIA A100 chips to run. According to market research firm Jibang Consulting,Most of the more than 1.2 million AI servers shipped this year are equipped with NVIDIA GPU chips, accounting for 60% to 70% of the market share.
Zhang Xiaorong pointed out to "Chinese Entrepreneur" that the latest US ban measures will have a huge impact on China's large model ecosystem: "China's large models rely on NVIDIA's GPU for training, and further bans will undoubtedly limit the technological breakthroughs and development of China's large models, affecting their research and innovation capabilities
For renting Nvidia's computing power cloud and cloud computing facilities of some domestic server companies, whether to place orders and stock up before the window period will become a major challenge. However, there are media reports that large companies such as Tencent and Baidu may be limited by the "ban" due to their "ample stock hoarding". Some manufacturers have also stated that although they have already stockpiled enough, there will still be great pressure in the future.
Domestic server manufacturer Xinhua San told China Entrepreneur: Regarding recent market rumors, we are closely monitoring and communicating with relevant parties to verify and evaluate potential impacts. Currently, the group is actively responding, and all business activities are still proceeding in an orderly manner. Xinhua San has always adhered to business ethics, strictly followed international business practices and laws and regulations, and conducted related business in compliance. In response to this incident, we have relevant plans in place and will actively communicate with relevant departments to continue to do so Follow up and follow up on the subsequent development of the event
After the news of the US ban was released,Chinese chip manufacturer Cambrian's stock price rose by over 6% in response.Industry insiders believe that the new ban in the United States has little impact on domestic AI chips produced by non American manufacturers, and there is a possibility that a group of emerging forces may emerge in the domestic computing power field.
Currently,The computational power of domestic large-scale model training is roughly divided into three parts,One is the computing power solution of Huawei Kunpeng and Shengteng AI ecology, without the participation of NVIDIA GPU; The second is hybrid computing power support, with a large number of NVIDIA A100 chips used, and some environments adding AMD and Intel chips, as well as domestic chips such as Cambrian and Haiguang, and accelerator card fusion running model training; The third is to rent servers with higher cost-effectiveness and cloud computing power to supplement the shortage of computing power.
However, chip development is a long-term and heavily invested task, and it is almost impossible for startup companies to quickly develop high-end GPUs comparable to NVIDIA in the short term. Apart from the technological gap, domestic GPUs are still not as well-established as NVIDIA in terms of ecological establishment.
News Hotline& Submission email: [email protected]
Tag: Trillion Nvidia evaporated billion overnight
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.