Source of the Article: “Science and Technology Board Daily”

Reporter: Wu Xuguang

The secondary market private placements of "big models" continue to be hot, and related regulatory inquiries have followed.

The "Science and Technology Board Daily" noticed that on September 21st alone, HaiTianRuiSheng and XingHuan Technology, two companies listed on the Science and Technology Innovation Board, disclosed announcements replying to inquiries.

From the questions and replies in the inquiry letter, including whether it involves duplicate construction, the role of the project in the issuer's continuous operation, and the market prospects related to the fundraising project, it can be seen that the issues related to the fundraising project market prospects have been mentioned multiple times from the perspective of the exchange's focus.

Regulatory Attention to the Rationality of Fundraising Projects

On September 21st, HaiTianRuiSheng released an announcement in response to the inquiry letter from the Shanghai Stock Exchange regarding the private placement, providing specific explanations on the necessity of the fundraising project, the scale of financing and benefit calculation, and the company's business operations.

In response to the question of whether the fundraising project involves duplicate investment, HaiTianRuiSheng replied that the raised funds are intended for the "AI big model training data set construction project" and the "data production vertical big model research and development project," and the investment projects are all centered around the company's current main business, expanding the company's existing product matrix and iterating and upgrading the existing technology system, without involving new products or technologies.

HaiTianRuiSheng announced a non-public offering of A-shares on June 21 this year, planning to raise no more than 790 million yuan for the AI big model training data set construction project and the data production vertical big model research and development project.

The private placement application of HaiTianRuiSheng was accepted on August 3 and received an inquiry letter from the Shanghai Stock Exchange on August 18. The Shanghai Stock Exchange requested the company to explain the differences and connections between the current fundraising project and the previous fundraising project, as well as the future commercial prospects of the current fundraising project and its impact on the issuer's continuous operation, in combination with industry development trends and industry competition.

In response, HaiTianRuiSheng stated in the inquiry reply letter that currently, domestic big models are mainly in the early development stage, and the downstream big model-related customer products are still in the early stage of first-generation product release/research and development, and the market has not yet been widely applied. The related data demand, especially the model capability optimization and transfer after market application, needs further release.

Zhang Zhe, General Manager of the Securities Affairs Department of HaiTianRuiSheng, told the reporter of "Science and Technology Board Daily," "The new wave of big models has opened up the ceiling of AI application, giving everyone a lot of room for imagination. The paradigm-level change brought about by big models is irreversible. If HaiTianRuiSheng can seize this opportunity window, it will be able to continue to maintain its market position over the next 5-10 years in the competition."

Not only HaiTianRuiSheng, but another company on the Science and Technology Innovation Board, XingHuan Technology, also received an inquiry letter from the Shanghai Stock Exchange on September 21 for the "big model" private placement, requiring the company to explain the differences and connections between the data analysis big model construction project and the existing business in terms of underlying technology, market positioning, product application scenarios, customer types, and sales models, and further explain whether the current fundraising is in line with the main business.

It should be pointed out that the data analysis big model construction project is a new product line developed by XingHuan Technology under the existing business.

XingHuan Technology stated that based on the implementation of the project, the company will focus on the core technology development of AI big models and conduct research and development work on big data big models, distributed vector databases, artificial intelligence infrastructure, and basic software series technologies. The data analysis big model construction project of the current fundraising project will be based on the company's technical accumulation in the big data field and the foundation of the previous data analysis products, focusing on the development and construction of big models in the vertical field of big data analysis, which is a new product line developed under the existing business.

XingHuan Technology emphasized that all the investment projects of the current fundraising have good market prospects, and through the implementation of the investment projects, new business income growth points can be cultivated and explored to enhance the company's profitability.

Behind the "Hot" Big Models

Under the AI boom, more and more companies are preparing to explore the research and application of big models from different perspectives, including technology giants such as Amazon, Google, Baidu, Huawei, Alibaba, and iFlytek, who have successively launched their own big model services.

Behind the "hot" big models, both the primary and secondary markets are considering whether the related investments injected can generate corresponding returns. Big models are indeed a new thing that requires long-term investment. Will it eventually become a conceptual investment?

Some private equity investors even bluntly stated that no matter which field they invest in, it must eventually be implemented, and relying on concepts is not sustainable.

There are also opinions from securities industry professionals holding the opposite view. They told the reporter of "Science and Technology Board Daily" that from the perspective of industry investment trends, the significant breakthroughs in artificial intelligence have had a major impact on social development, especially after the birth of AI big models led by ChatGPT, the marginal changes brought about by the core technologies of big models—deep learning and natural language processing technologies—are effective. It is intuitively visible that under the drive of the current AI hotspots, the valuations of industry-related listed companies have indeed increased significantly.

"From the perspective of industrial application landing, the progress of the relevant industries is not as fast as imagined, and it may take about two years for AI technology to sink into more application scenarios," the aforementioned securities industry professional believes.

It should be noted that current big model companies are all vying for commercial application opportunities, but they also face huge investments.

According to a research report by Guosheng Securities, the pre-training cost of a big model with 280 billion parameters is approximately $2 million per time, and Google's PaLM, which claims to have 540 billion parameters, will have a pre-training cost as high as $12 million per time.

The aforementioned securities industry professional further explained that the actions of XingHuan Technology and HaiTianRuiSheng are not isolated cases in the industry. Faced with high investment, private placements have basically become one of the most important ways for listed companies in the big model field to raise funds.

The "Science and Technology Board Daily" noticed that on the Science and Technology Innovation Board, Yuncong Technology and Tuolunsi have also successively disclosed private placement plans in the near future, intending to invest in big model research and development and other fields.

It is generally believed in the industry that the current investment cost of general big models is high and needs to be effectively combined with specific application scenarios. When the relevant industries are still in the early stage of development, how will companies balance the high investment in big models in terms of performance?

In response to this, Zhang Zhe, General Manager of the Securities Affairs Department of HaiTianRuiSheng, told the reporter of "Science and Technology Board Daily" that the two projects planned for this private placement refinancing, the "AI big model training data set construction project" and the "data production vertical big model research and development project," mainly focus on the direction of big model data solutions and involve large-scale research and development investment.

Zhang Zhe stated that according to the preliminary planning of the project, the "AI big model training data set construction project" is expected to become one of the new driving forces for income growth after its completion. The "data production vertical big model research and development project" will enhance the company's intelligent data processing capabilities under the big model paradigm, provide a strong technical foundation, and help the company achieve scale effects and reduce costs effectively.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。