Cloud fundraising from technology weighs heavily on AIGC, and the industry's large model is difficult to become a lifesaver

China Times (www.chinatimes.net.cn) reporter Lu Xiao and trainee reporter Shi Feiyue reported from Beijing

On the road of large models, cloud technology has the meaning of "dead knock" to the end. Following the launch of the basic large model "Calm Big Model", on July 21, the company announced that it will raise a total of no more than 3.635 billion yuan for industry large model projects, and pin one of its profit hopes on the future landing and commercialization of new projects.

Many industry leaders have indeed expressed optimism about the industry's big model, but it is still unknown whether Cloudwalk Technology can successfully raise the target amount in the short term when it has lost money for many years and its stock price has continued to fall. As for relying on the industry's large model to make profits, Zhang Xiaorong, president of the Deepin Science and Technology Research Institute, believes that at present, the maturity of domestic technology and the amount of data cannot make people see the hope of profit, and enterprises should focus on new technical routes and innovation activities.

Targeted fundraising of 3.6 billion yuan

According to the announcement of Cloudwalk Technology, the investment project raised this time is a large-scale model research and development project of Cloudwalk "industry wizard". Before the issuance and raised funds are in place, Cloudwalk Technology can invest in advance with self-financing funds according to the actual situation of the raised funds investment project, and replace them in accordance with the procedures stipulated by relevant laws and regulations after the raised funds are in place. After the funds raised in this offering are in place, if the actual net funds raised after deducting the issuance expenses are less than the total amount of funds to be invested in the fundraising, the insufficient part will be solved by the company's self-financing.

This fundraising exposed the current situation of insufficient funds of Cloudwalk Technology.

In May 2022, Cloudwalk Technology planned to raise 3.75 billion yuan through the issuance of shares (IPO) to the public, but the results were not satisfactory, and the final raised funds totaled 1.728 billion yuan, and after deducting relevant issuance expenses, the actual raised funds were 1.627 billion yuan. According to the latest announcement of Yuncong Technology, as of the end of 2022, the company has used a cumulative amount of raised funds of 1.25 billion yuan, accounting for 76.85% of the net funds raised.

At that time, Yuncong Technology also stated in the prospectus that if the actual raised funds (after deducting the issuance expenses) could not meet the investment needs of the fundraising project, the funding gap would be solved by the company through self-financing.

In the first year after listing, the 2022 financial report showed that Yuncong Technology not only suffered a sharp decline in revenue of 51.06%, but also lost 869 million yuan. From 2020 to 2022, Cloudwalk Technology has a cumulative loss of 2.314 billion yuan. In the first quarter of this year, Cloudwalk Technology continued to lose money, with a loss of 142 million yuan.

The funds raised this time are for the issuance of A-shares to specific targets, and the issuance targets are no more than 35 securities investment fund management companies, securities companies, trust and investment companies, financial companies, insurance institutional investors, qualified foreign institutional investors and other legal persons, natural persons or other qualified investors who meet the requirements stipulated by the CSRC.

But the final outcome is difficult to predict. Lens Research's founder Yuqing told the China Times that if you find a suitable investor and successfully issue it, you must also beware of variables in the secondary market, and if the stock price is too low at the time of issuance, you will not be able to raise the planned funds. "In addition, judging from the actual situation of Cloud Walk Technology's revenue decline and expansion of losses in recent years, it may not be possible or fail to issue because it is not favored by investors."

Favored by the AIGC concept, the stock price of Yuncong Technology ushered in a wave of gains this year. On January 3, Yuncong Technology's share price was 10.91 yuan / share, and the highest rose to 44.14 yuan / share on April 4, with a market value of four times that at the beginning of the year, but since then, Yuncong Technology, like some AI concept stocks, has begun to cool down, and as of the close of trading on July 21, its stock price has fallen to 16.43 yuan / share.

Bet on the big model of the industry

Different from the previously released basic large model "Calm Big Model", the "Industry Genius" large model belongs to the industry large model.

Industry large model refers to the large model optimized on the basis of the basic large model, for downstream tasks in specific fields, through supervised fine-tuning and reinforcement learning, etc., which can learn the specific knowledge and laws of the field or industry.

Since ChatGPT came out of the circle, major AI companies at home and abroad have launched large model products, some focusing on general large models, some focusing on industry large models, and the discussion of the two in the industry has not stopped.

In the view of cloud technology, the industry large model has the necessity of research and development in terms of cost and practicality.

In terms of cost, Yao Zhiqiang, co-founder of Yuncong Technology, said: "The cost of the iteration of the general large model is very high, although many traditional enterprises now have the impulse to try, slowly will calm down, independent development and operation iteration is really not cost-effective, this is the opportunity of the industry large model." ”

In terms of practicality, Cloudwalk Technology pointed out in the announcement that with the continuous evolution and development of the basic big model represented by GPT, the industry big model that can be realized through the downstream task migration of the basic big model will be an important measure for the future artificial intelligence platform to go deep into the industry.

Recently, Tang Daosheng, senior executive vice president of Tencent Group and CEO of the Cloud and Smart Industry Business Group, also affirmed the industry big model, "The general big model has strong capabilities, but it cannot solve the specific problems of many enterprises, and can solve 70%-80% of the problems in 100 scenarios, but may not be able to meet 100% of the needs of an enterprise in a certain scenario." However, if enterprises are based on the big industry model and fine-tuned by their own data, they can build exclusive models and create highly available intelligent services. ”

In the case of continuous losses, Cloudwalk Technology regards the general large model as one of the golden keys to future profits. "Although the implementation of this fundraising project will put some pressure on the company's profitability in the short term, it will have a positive role in promoting the company's future business development and profitability improvement in the medium and long term." Cloudwalk Technology wrote in the announcement.

The premise of profit is landing and commercialization, and the reporter of the "China Times" interviewed its relevant person in charge on the current application of Yuncong Technology's large model, and the other party did not disclose it, only said that the announcement prevailed.

However, in Zhang Xiaorong's view, betting on a big model is no longer a wise decision. "Domestic large models have limited technological maturity, there are inputs and no outputs, betting on large models, it is difficult to bring direct benefits to enterprises, at the same time, large models require a huge amount of capital and a huge amount of high-quality data, giant companies are often more difficult, small and medium-sized enterprises are more difficult to see the prospects."

Zhang Xiaorong believes that the development of the AI industry is a wave forward, the trough is soon after the trough, and there will often be a new trough after the trough, at the climax, the follow-up project rarely survives, and in the trough, there are often new companies or individuals inventing new technical routes, thus allowing the industry to enter a new stage of development.

"The development of large models is about to hit the ceiling, and this round of industrial boom is coming to an end, and resources and energy must be focused on new technical routes and new innovation activities in order to meet the opening of the next round of climax." Zhang Xiaorong said.