Over the past six months, more than 38 AI companies "bleed" to go to Hong Kong for IPO

According to the Titanium Media App, in the first half of 2023, there will be more than 38 AI or digital and intelligence-related companies queuing up on the Hong Kong Stock Exchange, accounting for about 40% of the IPOs of companies queuing up on the Hong Kong Stock Exchange.

Author｜Editor Lin Zhijia｜Li Xiaonian

Source: Titanium Media

As ChatGPT and artificial intelligence (AI) large-scale models sparked a global upsurge, the capital market's view of AI companies has warmed up, and many star AI companies have gone public collectively.

Titanium Media App reported on July 13 that AI pharmaceutical company InSilico Medicine Cayman TopCo (hereinafter referred to as "InSilico Intelligence") recently submitted a listing application to the Hong Kong Stock Exchange. The sponsors of the company are Morgan Stanley, CICC, and BNP Paribas will jointly assume the responsibility of global general coordination.

The IPO launch of Insilicon Intelligence has aroused industry attention. On the one hand, it is because Insilicon Intelligence is the first AI pharmaceutical company to launch an IPO in the Asia-Pacific region, and is expected to become "the first AI pharmaceutical company in the Asia-Pacific"; on the other hand, Insilicon Intelligence has not achieved new drug volume Production and sales, the revenue in 2022 is only 30.147 million US dollars, the loss during the year is as high as 221 million US dollars, and the adjusted loss is 67.577 million US dollars-similar to other AI companies, there are short-term losses and difficulties in making profits.

Of course, it's not just Yingsi Intelligence. Recently, many AI companies, such as AI smart driving chip company "Black Sesame Smart", AI voice company "Yunzhisheng", generative AI (AIGC) company Going out to ask, etc., have intensively submitted listing applications to the Hong Kong Stock Exchange, waiting in line Approval for listing.

According to Titanium Media App, in the first half of 2023, there will be more than 38 AI or digital and intelligence-related companies queuing up on the Hong Kong Stock Exchange, accounting for about 40% of the 96 IPO companies currently being processed on the Hong Kong Stock Exchange.

Yingsi lost 780 million in two years, Yunzhisheng lost 560 million in three years, and Heizhima Smart lost 1.58 billion

At the end of June, before the semi-annual financial update, three star companies, iSilicon, Yunzhisheng and Heizhima, submitted their prospectuses. These three are all star companies in the AI industry.

The first is Yingsi Intelligence.

Insilicon Intelligence was founded in 2014 by Dr. Alex Zhavoronkov. It is currently headquartered in Hong Kong, China. Dr. Alex is the chairman of the board of directors, executive director and CEO of the company. His nationality is Canadian and Latvian. He graduated from Queen's University in Canada.

In terms of business, Insilicon Intelligence is a world-leading end-to-end biotechnology company driven by generative AI in the AIDD (artificial intelligence-assisted drug discovery) pipeline development process. , chemical and clinical development, to accelerate the discovery and development of new drugs.

In 2021, in less than 18 months, the Insilicon intelligent team used the self-developed AI drug research and development platform to complete the entire process from conception to preclinical candidate compound discovery, discovered new targets for pan-fibrosis, and targeted This target designs a preclinical candidate drug for the treatment of idiopathic pulmonary fibrosis (IPF), which has been proven to have significant efficacy in preclinical in vivo and in vitro experiments, and the cost is much lower than the investment of conventional drug discovery. (For details, see the previous article: "Why did AI become the "medicine god"? | Titanium Media Depth")

Alex previously told Titanium Media App that in his opinion, although the AI pharmaceutical industry will not explode overnight, it may take more than 10 or 15 years. But from the perspective of commercialization, he believes that this track mainly has three commercialization models: AIDD pipeline promotion, project development cooperation, and software licensing cooperation.

In terms of products, the prospectus shows that the company's core product is a potent selective inhibitor of TNIK, ISM001-055, which has a high affinity and has the potential to treat IPF. The results of Phase I clinical trials completed in New Zealand show that ISM001-055 has good safety, tolerance and PK data on healthy volunteers. The results of the Phase I clinical trial completed in China also showed that it has similar good safety and tolerability in healthy volunteers. Insilicon Intelligence launched the Phase IIa clinical trial in China under the one-time approval process of NMPA in April 2023. In February 2023, Insilicon Intelligence submitted an IND application for the US Phase IIa trial to the US FDA, and obtained the FDA's IND approval in June 2023, and plans to start Phase IIb and Phase III studies.

In addition, ISM001-055 received Orphan Drug Designation from the FDA in February 2023, qualifying the company for incentives, including potential seven-year market exclusivity upon approval.

Since 2018, Insilicon has gone through 7 rounds of financing, raising a total of US$407.5 million. Investors include Sequoia China, Qiming Venture Partners, Lilly Asia Ventures, WuXi AppTec, Sinovation Ventures, Baidu Ventures, Hillhouse , Warburg Pincus, Fosun Pharma and other institutions. From 2018 to 2022, the post-investment valuation of Insilicon Intelligence rose from US$54.4 million to US$895 million, a four-year increase of nearly 16 times.

From the perspective of performance, Yingsi Intelligence is in a state of poor profitability and long-term losses: In 2021 and 2022, Yingsi Intelligence’s revenue will be US$4.713 million and US$30.147 million, an increase of 539.6%; the total loss and comprehensive expenses during the year are 131 million US$, US$221 million; adjusted losses were US$42.858 million and US$67.577 million respectively. The total loss and comprehensive expenses in the two years totaled 352 million US dollars (about 2.523 billion yuan), and the adjusted loss totaled 110 million US dollars (about 780 million yuan).

There are two main reasons for this: first, the accounting method of the financial data of Hong Kong stocks is different from that of A shares and Hong Kong stocks, especially to add some employee and equity incentive expenditure data, which shows that the overall loss is relatively large; second, the company’s income is only drug research and development. There are two types of cooperation services and software solution services, which are highly dependent on the drug R&D cooperation model with enterprises. The software solution revenue growth is slow, and while there is no mass production and sales of new drugs, a large amount of R&D expenditure is required, resulting in a state of loss.

The prospectus shows that in 2021 and 2022, the net cash used by Insilicon Smart for business activities will be US$38.070 million and US$47.517 million, respectively.

As of the Latest Practicable Date, Insilicon Intelligence's pipeline included one clinical stage asset in phase II study, two clinical stage assets in phase I study, ten preclinical assets and multiple early discovery stage assets.

In addition, Insilicon is different from other innovative drug research and development companies. In fact, there are US stock reference companies as the benchmarks for AI new drug research and development—Schrodinger, Exscientia (NASDAQ: EXAI), etc. They are in a similar situation to Insilicon. It is selling software and cooperative services, especially software with a high gross profit margin.

The financial report shows that in 2022, Schrödinger’s total revenue will be US$181 million, an increase of 31% compared to US$137.9 million in 2021; of which, the annual software revenue will be US$135.6 million, a year-on-year increase of 20%, and the gross profit margin will be 78%. The loss was $149.2 million, widening from a loss of $101 million in 2021. In addition, during the same period, Exscientia's total annual revenue was US$32.9 million, a decrease from 2021, and R&D investment was as high as US$53.2 million, which was 161% of total revenue, while the net loss reached US$143 million.

Therefore, on the whole, the business of Insilicon Intelligence is at a similar level to that of Schrödinger and Exscientia. The difference is that Schrödinger relies on software services, while Insilicon Intelligence needs the support of partners, and Exscientia relies on experimental cooperation and gene sequencing business income. In addition, the stock prices of many AI pharmaceutical companies continue to fall, with an average drop of about 78% from the highest point, which poses certain challenges for Insilicon Intelligence.

According to the prospectus, Yingsi Intelligent’s listing in Hong Kong intends to use the raised funds for the following purposes: to provide funds for further clinical research and development of the core product ISM001-055; to provide funds for the research and development of other pipeline drug candidates; for further development and the expansion of the robotics laboratory; for the development of next-generation artificial intelligence models and related verification work; and will be used for working capital and other general corporate purposes.

The second is Yunzhisheng, an AI voice company that has moved from the Science and Technology Innovation Board to Hong Kong stocks.

This time, Yunzhisheng intends to list on the main board of Hong Kong, and the joint sponsors are CICC and Haitong International.

The company's mission has changed from the previous direction of AI voice to "creating a world of interconnected intuition through general artificial intelligence (AGI)", that is to say, Yunzhisheng hopes to establish a business model in the direction of AGI.

Yunzhisheng said that the company has the greatest potential for AGI commercialization. Shortly after the Transformer algorithm in 2017 and the BERT algorithm made breakthroughs in natural language processing in 2018, we launched the first BERT-based large language model UniCore (Shanhai) and related AI solutions for customers in vertical industries by using our technology in interactive AI. The company's AI technology continues to develop, with multi-modal capabilities and a large number of dynamic user interaction iterations in different application scenarios.

In terms of performance, the prospectus shows that from 2020 to 2022, Yunzhisheng's operating income will be 261 million yuan, 456 million yuan and 601 million yuan respectively, and the year-on-year growth rates will be 19.65%, 75.29% and 31.33%, respectively. The growth rate is relatively fast However, the overall revenue scale is still very small; the annual losses in the same period were 298 million yuan, 434 million yuan and 375 million yuan respectively, and the accumulated losses in the past three years have exceeded 1.1 billion yuan, and the loss range in 2022 will be relatively narrowed; the adjusted net loss for the year They were 210 million yuan, 172 million yuan and 183 million yuan respectively, with a cumulative loss of 560 million yuan in three years.

Yunzhisheng spends a lot on R&D. From 2020 to 2022, the company's annual research and development expenses will be 189 million yuan, 188 million yuan and 287 million yuan respectively, accounting for 72.6%, 41.3% and 47.8% of the total revenue during the same period.

In terms of financing, from April to June 2023, Yunzhisheng completed the D3 round of financing, with a transaction amount of 700 million yuan. Up to now, Unisound investors include Qiming Venture Partners, China Internet Investment, JD.com, Qualcomm Ventures and other institutions.

In late May this year, UniSound launched its own large language model, the Shanhai Large Model, with 60 billion parameters. Huang Wei, the founder of the company, said that Shanhai’s goal is to match the general capabilities of ChatGPT based on GPT-3.5 in 2023, and surpass GPT-4 in multiple vertical fields such as medical care, IoT, and education.

In terms of business model, Yunzhisheng’s revenue sources mainly include smart life and smart medical care. Last year, the two business revenues accounted for 81% and 18.9% respectively.

Judging from the prospectus, Yunzhisheng's top five customers have undergone major changes, high customer concentration, unstable customer groups, and the company's payment collection cycle is long. In the past three years, the net cash flow has continued to flow out, which was -166 million yuan last year. As of the end of last year, the company's cash and equivalents were only 74 million yuan. By April this year, with the partial recovery of receivables, it reached 193 million yuan.

“We believe that future revenue growth will depend, among other things, on our ability to develop new technologies, successfully commercialize AI technology, enhance customer experience, compete effectively, and expand our customer base. In addition, we expect the company to incur substantial costs and Expenses. If we are unable to generate sufficient revenue and manage expenses, we may continue to incur losses in the future and may not be able to achieve or subsequently maintain profitability." Yunzhisheng stated in the prospectus.

This time, UniSound’s listing in Hong Kong intends to use the raised funds for the following purposes: investment in Atlas artificial intelligence infrastructure, Yunzhisheng Brain upgrade, talent training and joint research and development, international expansion and strategic cooperation, and general purposes.

Finally, it is the first technology innovation company to submit a prospectus in accordance with the 18C rules of the Hong Kong Stock Exchange, the intelligent driving computing chip company Heizhima Smart.

In order to increase the attractiveness of special technology companies to list in Hong Kong, the Hong Kong Stock Exchange added Chapter 18C to the "Main Board Listing Rules" on March 31, 2023, and launched a listing mechanism for special technology companies, allowing no-income, Unprofitable technology companies listed in Hong Kong. The new rules apply to companies in the five specialized technology industries of new generation information technology, advanced hardware, advanced materials, new energy, energy conservation and environmental protection, new food and agricultural technology.

Founded in July 2016, Black Sesame Smart is a supplier of car-grade smart car computing SoC chip solutions. The company's own car-grade products and technologies, including autonomous driving, smart cockpit, advanced imaging and interconnection, etc., are strategically giving priority to the development and commercialization of L2 to L3 smart driving chip products at this stage, and have successively launched Huashan series A1000 and A1000L , A1000 Pro and Wudang series C1200 car-grade SoC.

According to Frost & Sullivan's data, in terms of shipments of automotive-grade high-computing SoCs in 2022, Black Sesame Smart is the world's third largest supplier, with a global market share and a Chinese market share of 4.8% and 5.2%, second only to Nvidia and Horizon.

In terms of financing, since its establishment, Black Sesame Smart has completed 10 rounds of external financing. The company’s valuation exceeds US$2.218 billion (approximately RMB 15.895 billion). Investors include Weilai Capital, Northern Light Venture Capital, Junhai Chuangxin, SAIC Group, SK China, China Merchants Group, Xiaomi Changjiang Industrial Fund, Wuyuefeng Capital, Legend Venture Capital, Linxin Capital, etc.

However, judging from the Black Sesame Smart Prospectus, although it is in a leading position in the field of autonomous driving, it is still in a state of huge losses and high R&D investment.

From 2020 to 2022, Black Sesame Smart achieved revenues of 53.021 million yuan, 60.504 million yuan, and 165 million yuan; the net losses during the year were 760 million yuan, 2.357 billion yuan, and 2.754 billion yuan, with a total loss of 5.871 billion yuan; The adjusted net losses were 273 million yuan, 614 million yuan, and 700 million yuan, and the three-year adjusted loss reached 1.587 billion yuan; at the same time, the research and development investment was 255 million yuan, 595 million yuan, and 764 million yuan, up to 1.6 About 100 million yuan.

In terms of specific business, Black Sesame Intelligence mainly relies on autonomous driving products and solutions, and intelligent imaging solutions. Among them, in 2022, the smart driving business revenue will account for as high as 86%, an increase from 56.6% in 2021, mainly due to the increase in sales of its products and solutions and the mass production of its own SoC by the end of 2022.

As of now, Black Sesame Smart has received intentional orders for 15 models from 10 car OEMs and first-tier suppliers, and has accumulated more than 30 cooperative users, including FAW Group, Dongfeng Group, Geely Group, JAC Group, Bosch Waiting for the company. As of 2022, the customer base of Black Sesame Smart has grown from 33 at the end of 2020 to 89.

In addition, Titan Media App combed through the prospectus and noticed that the income of Black Sesame Smart depends on large customers. According to the prospectus, in 2022, Black Sesame Smart's revenue from the five largest customers will account for 75.4% of the total revenue, of which the revenue from the largest customer will account for 43.5% of the current period. Black Sesame Intelligence stated that if the company loses one or more major customers or a significant reduction in sales in the future, it may have an adverse impact on the business, operating performance and financial status of Black Sesame Intelligence.

If successfully listed, Black Sesame Smart is expected to become the "first stock of domestic self-driving chips" in Hong Kong stocks.

This time, the company’s listing in Hong Kong plans to raise funds of 200 million to 300 million US dollars, which will be mainly used for research and development in the next three years: 80% will be invested in the research and development of smart car-grade SoC, smart car software development, and autonomous driving solutions; 10 % to improve the company's commercialization capabilities; 10% is used for working capital and general company purposes, especially the purchase of inventory for mass production of SoC chips.

"Hematopoiesis is difficult", AI company "Qian" Tu Molu

Since the end of November last year, ChatGPT has triggered a new wave of global AI. The capital market has poured into ChatGPT and AI large-scale model concept stocks, making the stock prices of many AI companies such as Cambrian and Haitian AAC rise steadily. Since the beginning of the year, the Cambrian has risen by 245.70%, Haitian Ruisheng has risen by 94.57%, and HKUST Xunfei has risen by 98.28%.

At the same time, this has also allowed many unlisted AI and cloud computing companies to see the best listing approval window and dawn, and more than 38 companies have poured into the Hong Kong Stock Exchange.

However, according to the analysis of Titanium Media App, the three AI companies mentioned above, iSilicon, Yunzhisheng and Heizhima, face common performance challenges: low revenue, long-term large losses, R&D failure, profitability and " "Hematopoietic" ability is poor, and investors urgently need to exit and realize.

For example, Black Sesame Smart has an annual revenue of 165 million yuan, but research and development is more than seven times the revenue, reaching 764 million yuan. Coupled with "tape-out" and sales expenses, the company's loss rate remains high.

The long-term "blood-making" ability of AI companies is still an important challenge facing the entire industry.

Wang Bing, a partner of Oriental Fortune, believes that China's environment in the field of software and algorithms is very different from that of the United States. "We have seen several big gaps. The first is the company. In fact, large Chinese companies have no boundaries. I will copy whoever does well, regardless of To B or To C. Basically, there is nothing they can't do. Son. So for small companies, entrepreneurs who do algorithms or software need to be 'stuck' to a point where either big companies can't do it, or big companies won't do it, but the possibility of not being able to do it is not particularly high .Start-up companies need to find some resources or barriers, which are places where real big companies cannot enter.”

Liu Di, managing director of Shengjing Jiacheng, told Titanium Media App that at present, the bottom layer of the big AI model is mainly data, computing power, and algorithms, and the core technology is less difficult—computing power needs to be stacked with Nvidia graphics cards, and algorithms can also use open source. There are regulatory restrictions, so investors' interest is reduced. "This means that everyone's starting line is the same, so the level gap is not big."

Wang Xuhui, general manager of Huahui Chuangfu Investment, said that ChatGPT is currently in the stage of conceptual speculation, and has little short-term impact on the operation and profitability of related companies. However, AIGC (generative AI) has a long track and a large market space. Recently, the market sentiment has become more positive and optimistic, which is more beneficial to the growth direction of technology including AIGC. However, there are great uncertainties in the development of AI technology. Investors should moderately avoid the volatility risks of related concept stocks and grasp the investment rhythm.

Recently, whether it is data collection, analysis and computing power companies in the upstream of the AI industry chain, or AI software companies in the middle reaches of the industry chain, stock prices have fallen. good condition.

On the evening of July 11, Inspur Information released a performance forecast stating that the company expects to achieve a net profit of 286 million to 382 million yuan in the first half of the year, a year-on-year decrease of 60%-70%; 100 million yuan, a year-on-year decrease of 88%-99%; the 2023 semi-annual performance forecast released by HKUST Xunfei on the evening of July 10 shows that the company expects the net profit attributable to the parent in the first half of the year to be 55 million to 80 million yuan, a decrease compared with the same period last year. 71% to 80%; and the estimated loss after deducting non-net profit is 270 million to 330 million yuan.

At present, the entire AI industry is ushering in a new upsurge. There is a certain degree of uncertainty about who will win the big model, but it is basically "burning money", and continuous investment is difficult to make a profit. Whether it is upstream sales The wave information of the server "shovel", the Cambrian, and iFLYTEK, which sells algorithms and software in the midstream, have all proved this point.

Although the large model has promoted AI technology to enter a new stage of development, it may become the development engine of society and a new productivity of human beings, which can greatly benefit society and greatly promote development. But AI, like anything else, may also bring about negative impacts. The collection of personal information and the protection of personal privacy are more challenging. Policies will continue to regulate the AI industry, and the company itself is constantly adapting to the environment. In the long run, the profitability of the AI industry still faces huge uncertainties, which requires the market to think "calmly".

(This article is the first release of Titanium Media App)