Riding the tide of AI, the fourth paradigm is about to go on the market.
From September 18 to September 21, Fourth Paradigm will conduct external listings and plan to be listed on the Hong Kong Stock Exchange on September 28.
Fourth Paradigm was founded in 2014 by Dai Wenyuan (Dai Wenyuan was responsible for the R&D and management of Baidu's search advertising system "Phoenix Nest").
The company name Fourth Paradigm comes from a speech by Turing Award winner Jim Gray, who summarized the scientific development process into four "paradigms":
The first paradigm is experimental, which is mainly based on recording and describing natural phenomena;
The second paradigm is theory, which uses models to summarize phenomena recorded in the past;
The third normal form is computation, which solves complex problems by designing algorithms and writing programs to run on computers;
The fourth paradigm is data, by collecting a large amount of data, let the computer summarize the rules.
This name is very consistent with the business of Fourth Paradigm, which is currently the largest platform-centric decision-making artificial intelligence company in China, according to the prospectus.
As we all know, as long as the determiner is added long enough, then anyone can be the first place.
So, what does the artificial intelligence company of Fourth Paradigm do?
According to the different application fields, the artificial intelligence industry can be divided into four categories:
(1) Visual artificial intelligence, based on visual data to identify, track and measure objects, and convert information into insight and judgment, the main application is security, we are familiar with the companies such as Megvii Technology and YITU Technology;
(2) Speech and semantic artificial intelligence, recognition, generation and exchange of speech, text and other language information, the main application is voice input method, intelligent customer service, etc., we are familiar with the company iFLYTEK;
(3) Artificial intelligence robots, designed to replace humans to perform certain repetitive or dangerous tasks, the main applications are industrial drones, surgical robots, etc., we are familiar with listed companies such as intuitive surgery in the United States, minimally invasive robots in China, etc.;
(4) Decision-making artificial intelligence, that is, the work of the fourth paradigm, is to help the decision-making process by identifying hidden patterns in data, and specific applications include loan risk identification and anti-money laundering.
According to the prospectus, the fourth paradigm has been implemented in finance, retail, manufacturing, energy and power, telecommunications, transportation, technology, education, media, healthcare and other scenarios.
According to the China Insights Consulting report, among the above four types of artificial intelligence applications, the market size of decision-making artificial intelligence is the fastest growing. In 2022, the spending scale of China's decision-making AI market was 53.2 billion yuan, and it is expected to grow to 210.4 billion yuan in 2027, with an average annual compound growth rate of 31.7%.
The fourth paradigm does eat the cake too. In 2020, 2021 and 2022, the revenue of Fourth Paradigm was 942 million yuan, 2.018 billion yuan and 3.083 billion yuan, respectively, an increase of 105.0%, 114.2% and 52.7% year-on-year.
During the period, the number of users of Fourth Paradigm increased from 156 in 2020 to 409 in 2022, the number of benchmark users (Fortune Global 500 companies or listed companies) increased from 47 to 104 during the year, and the average revenue per benchmark user increased from 12.3 million yuan to 17.9 million yuan.
Although it is high-tech and high-growth, in fact, the money of the fourth paradigm is not easy to earn.
First, as revenue grows, so do the accounts receivable of the fourth paradigm. Receivables in 2020, 2021 and 2022 were $263 million, $778 million and $1,493 million, accounting for 27.9%, 38.5% and 48.4% of revenue, respectively.
This means that the revenue of the fourth paradigm in 2022 is almost in exchange for a "free trial for 6 months", and the quality of this revenue growth is not high.
Specifically, the revenue of the fourth paradigm is mainly divided into two parts, one is the "prophet platform" and products, and the other is application development and other services.
The Prophet platform, simply put, is a general AI model, and after the platform is developed, it can be directly sold to different customers, and there is no need to customize it later. Develop once, keep making money.
Application development and other services are to assign employees to develop projects separately according to the different characteristics and needs of customers, and each project must be redeveloped, and only one project will be earned (service fees will be charged for continuous use).
Compared with the two, the revenue of the Prophet platform is highly replicable, it is multiplication, and the revenue = 1 platform × sales. And project-based income, is to do addition, income = 1 project + 1 project + 1 project..., the expansion speed and stability are much worse.
Unfortunately, the platform revenue of the fourth paradigm has not risen much, and the proportion of project-based revenue has increased, from 34.3% in 2020 to 57.8% in the first 3 months of this year.
From the perspective of profit, from 2020 to 2022, the operating losses of the fourth paradigm are 560 million yuan, 1.17 billion yuan and 1.03 billion yuan, respectively.
The cost of sales alone (the cost of hardware components, technical service fees, employee welfare expenses) and R&D expenses add up to more than revenue, so the loss of the fourth paradigm is not a solution.
This shows that such non-general-purpose types of AI services still do not sell for a price.
If the cost of share-based incentives and the interest expense of redemption liabilities are added back (related to the share repurchase agreement and do not need to be paid after successful listing), the loss of the fourth paradigm will be significantly narrowed, and the loss will be reduced to 500 million yuan in 2022.
However, the 430 million yuan of equity incentives is actually impossible to be less - go to large companies to engage in research and development, there are high salaries, there are also equity, if the fourth paradigm does not give a large amount of equity, employees have no reason to work in the fourth paradigm.
As for the "equity inflation" caused by the large issuance of equity incentives, that is, profit dilution, it is up to each shareholder to bear it.
Because of its inability to make blood, Fourth Paradigm has experienced at least 11 financings in just 8 years since its inception in 2014. Of course, this can also be explained by the fact that investment institutions continue to be optimistic about the development prospects of the fourth paradigm.
So, although the use of decision-making AI can obviously be applied to all walks of life, it is promising and has a future.
However, it is certain that the fourth paradigm of landing on the capital market will still require the generosity of shareholders in the short term.
After all, the word "future" is read as future, written as future, but its precise interpretation should be:
Burn money, burn a future.