In 2023, the "barbarians" in the computing power market are menacing.
In 2024, the "barbarian" trend of the computing power market will be embarrassing.
Last year, "computing power" was pushed to the hottest position in history, and the capital market was known and respected by both institutions and retail investors. Just like real estate and blockchain in the past, today's automobiles, there are always people who sniff money, enter the game in a high-profile manner, and collide across borders.
Printing lottery tickets are rapidly resold to computing power and create prosperity
In the first half of 2023, A-shares set off a wave of hype about the concept of AIGC, and a company named Hongbo Co., Ltd. rose by 427.52% in the first half of the year, ranking fifth among A-share listed companies. How did this listed company, which had fallen for six consecutive years, achieve such results?
Hongbo shares publicly disclosed main business,Lottery printing, sales, packaging and printing, etc.,Proper traditional industry,And the lottery printing industry has limited space,With the decline in lottery popularity in recent years,Business performance is getting worse.
In June 2023, Hongbo quietly established a subsidiary, InBev Digital, to enter the computing power industry. At that time, Hongbo shares rose to the limit for two consecutive trading days, and the stock price jumped from the lowest point of 4.54 yuan to 6.75 yuan within a month, and then the stock price once rushed to the highest 45.29 yuan, and the market value increased 10 times.
The reason is that the company has "some" cooperative relations with NVIDIA, and at that time, the United States tightened sanctions on China's AI computing chips, and NVIDIA's accelerator cards have become a scarce resource, and product prices are getting higher and higher day by day.
Only two months later, Zhongguancun Zhongheng Cultural and Technological Innovation Service Alliance, NVIDIA, and InBev Digital Technology jointly established the Beijing AI Innovation Empowerment Center in Beijing, InBev Digital is responsible for the infrastructure part of the intelligent computing center, and the profit model is to provide public computing power for rent, and the first phase is composed of a standard unit supercomputing server (20 NVIDIA 80GB DGX servers), which can provide 100P computing power.
According to the market price at that time, the purchase price of 20 NVIDIA 80GB DGX servers in the first phase was 4-50 million yuan. According to the public interview information of Zhou Weiwei, then the head of InBev Digital and the deputy general manager of Hongbo Co., Ltd., it is expected that by the end of 2023, the scale of InBev Digital's own computing power will be expanded to 3000P. To this end, Hongbo shares also need to prepare at least 3-350 million yuan of funds.
Huge capital investment in the popular computing power track seems to create a win-win feast for capital, technology, and industry, but the reality is not like this.
The distribution of benefits is controversial, and the mess is unattended
Just last week, it was reported that Zhou Weiwei, the head of InBev Digital, a wholly-owned subsidiary of Hongbo Co., Ltd., and others have been dismissed. On April 16, InBev Digital announced that it had dismissed some employees from their positions in accordance with relevant regulations, but did not mention the specific names of those who were dismissed.
Previously, Hongbo Co., Ltd., the parent company of InBev Digital, attracted public attention due to its performance change. According to the announcement of Hongbo shares, the estimated net profit in 2023 will be lowered from 37.4 million yuan to 56.1 million yuan to an estimated loss of 50 million yuan to 58 million yuan.
Analysts believe that a contract payment of 500 million yuan from InBev Digital may be the key to the "change of face" of Hongbo's performance. The performance reversal is one of the "fuses" of the executive turmoil.
After the stock price rose sharply in 2023, although the relevant executives of Hongbo Co., Ltd. made a commitment not to reduce their holdings, it is too coincidental that through several equity changes and the "judicial means" of contract disputes, several core shareholders and executives have essentially completed the "passive reduction", and there is not much equity left. Under repeated operations, Hongbo shares have now become a company without actual controllers.
Not only that, the senior personnel of Hongbo shares have also changed many times this year. On January 14, 2024, more than 9 months after the resignation of former chairman Mao Wei, the former director Ni Hui was elected as the chairman of the company. On January 15, Li Honglei, general manager of Hongbo Co., Ltd., resigned, but was reappointed as executive deputy general manager. On the same day, You Qingquan, the company's securities affairs representative, resigned. and is fully responsible for the computing power sector, and Zhou Weiwei, a "beautiful executive" who was once praised by Hongbo, was also dismissed.
Computing power construction is not a capital game, and the "barbarians" will eventually retire
In the past two years, computing power, as an important productive force in the development of the digital economy, has been paid more and more attention, and both national policies and industrial investment have been significantly improved. This has also attracted the active participation of many market players.
However, fundamentally speaking, the field of computing power is a typical technology-intensive industry, and the entire industry has a high professional threshold and technical barriers, and it is not a simple capital-driven market.
In recent years, the hot intelligent computing center has become a specific landing carrier for computing power to serve the real industry and digital economy, and it is also becoming the main force and highlight of local new infrastructure.
Due to the scarcity and premium brought about by Nvidia's ban last year, a large number of outsiders suddenly entered the field of intelligent computing centers, trying to "make a profit" quickly and roughly.
In addition to Hongbo shares, there are also Lotus to do monosodium glutamate, Mercedes-Benz Information to do electronic reading, Hengrun shares to do machinery manufacturing, and ST Ruide to make sapphire materials. People who look at various cross-border forces are dumbfounded, no matter what background they used to have, it seems that as long as they have money, they can enjoy themselves in the computing power market.
Without exception, these companies have obtained NVIDIA's in-demand cards or related resources because they have received the blessing of certain channels or computing power brokers, and some do not even have their own data centers, so they are rushing to the market to make money. Some industry insiders broke the news that Hongbo relied on leasing other people's computer rooms to do business, and it looked ugly.
Similar behaviors were not uncommon last year, most of them had similar routines and the same goal: to set up a subsidiary, enter the computing center market through announcements, and quickly raise the stock price; As for whether it is really operating a computing power center, whether it has core computing power technology, and whether it can indeed provide computing power services, there is little attention.
Unexpectedly, the script reversed too quickly. Hongbo shares' continuous equity changes, illegal passive reductions, management changes, and performance downward revisions have poured cold water on the impetuous market. It turns out that you only know who is swimming naked at low tide, and only when you have a sharp eye can you see who is a "monster".
The intelligent computing center is neither a "card" sales nor a pile of equipment, but a professional public computing infrastructure with social and industrial value, which needs to be considered and built from multiple dimensions such as technical performance, industrial support, application adaptation, and operation services.
At present, there are three obvious misunderstandings in the construction of intelligent computing centers.
The first is the single pursuit of computing power scale, although the state encourages "moderately advanced construction", but it cannot ignore the adaptability and sustainable development of industrial applications;
the second is to over-pursue or over-promise short-term profit returns, ignoring its public service attributes and industrial cultivation and traction attributes as public infrastructure;
The third is to focus on construction and not operation, which will lead to a disconnect between design and planning and application, and it will not be used well, and it will not be able to play its due effect, resulting in a number of unfinished "chicken ribs" projects, and "integration of construction and operation" has long become an important standard in this field.
At this stage, due to the popularity of the computing power market, capital has entered on a large scale, most of which is to speculate the stock price with the help of the concept of computing power, or to make quick money as NVIDIA's "shovel seller", with the intention of quickly sharing the soup in the market.
Some enterprises dare to promise anything in order to get the project, but in fact they have no delivery ability at all, and even do not have all the qualifications; some only have "cards" and have no other technology and related equipment supply chain, but they promise to do system construction and operation services. In this case, there is no way to ensure the smooth implementation of the project, and even if it is barely delivered, there will be many problems, which will bury huge hidden dangers for subsequent operations.
It is the iron law of any industry that professionals do professional things. After the farce, it is time for the computing power market to return to normal.
Just like the rise of domestic computing power after the big wave of NVIDIA, this is an inevitable objective law of the market. The construction of computing power is related to the overall development of the digital economy, and it is still necessary to follow the rigorous model of market traction, government investment, and professional landing. The "barbarians" who simply want to "fry a wave and make a profit", although they have brought the industry hot in a short time, but without continuous technical cultivation and professional service capabilities, they will only disrupt and destroy the market in the end.