From Matrix in "The Matrix" to adult parks in "Westworld", artificial intelligence has long been developed to the level of thinking in film and television dramas.
In real life, as AlphaGo defeats the talented teenager of domestic Go, artificial intelligence has almost evolved to the stage of surpassing the human brain.
China is fully prepared for this intelligent battlefield, which is bound to cause giants to compete for hegemony. In 2017, the "New Generation Artificial Intelligence Development Plan" was introduced, officially elevating artificial intelligence to the height of "national strategy".
A number of AI companies are also rising up.
Today, A shares have ushered in the real sense of "artificial intelligence first stock" - Glin Deep Pupil.
Compared with the bumpy and difficult listing process of other AI companies in Hong Kong stocks and A-shares, Geling Deep Pupil has only experienced more than half a year from acceptance to meeting to listing, which is also smooth.
On March 7, the Geling Deep Pupil Science and Technology Innovation Board launched the subscription, the issue price is 39.49 yuan / share, the number of shares issued does not exceed 46.24 million shares, and the total amount of funds raised is expected to reach 1.826 billion yuan.
At the opening of the market on March 17, Geling Deep Pupil officially landed on the Science and Technology Innovation Board, the stock code 688207, and as of the close, its total market value was about 6.9 billion yuan.
When our eyes penetrate the prospectus, we can see its unique charm from a multi-dimensional perspective: not hurried, and firm in steps.
01
National strategy and spray thin market
If we look back at the history of human development from the perspective of technology, we will find that every time the productive forces are pushed to the next more advanced stage, it is actually hard technology, that is, the key core technology that has a major supporting role in economic and social development.
Quantitative historians have pointed out that the Industrial Revolution was the greatest thing in human history.
As the world's second largest economy, we have been striving to build a modern socialist power. However, in the development process of the industrial revolution in previous generations, China has been in a position of follow-up for a long time and has not been able to lead.
When we enter the era of Industry 4.0, China has entered the first echelon of the "Fourth Industrial Revolution" with artificial intelligence as the core, and regards artificial intelligence as the guidance of future strategies.
In 2016, the State Council issued the "13th Five-Year Plan" National Science and Technology Innovation Plan, which clearly regards artificial intelligence as the main direction for the development of a new generation of information technology.
In July 2017, the State Council promulgated the "New Generation Artificial Intelligence Development Plan", which officially raised the development of artificial intelligence to the level of national strategy.
On this basis, the Ministry of Industry and Information Technology issued the Action Plan for Promoting the Development of a New Generation of Artificial Intelligence Industry (2018-2020) in December 2018, which put forward specific action goals, focusing not only on basic theories and key technologies, but also on the implementation of various important industries.
In March 2021, the 14th Five-Year Plan adopted by the Fourth Session of the 13th National People's Congress clearly proposed to "cultivate and expand emerging digital industries such as artificial intelligence, big data, blockchain, cloud computing, and network security", creating a good policy environment for the sustainable and rapid development of the mainland artificial intelligence industry and the company in the future of the 14th Five-Year Plan or even a longer period of time.
In this context, the artificial intelligence industry has made great progress.
According to iResearch data, in 2019, the core industries and driving industries of artificial intelligence in mainland China were 108.86 billion yuan and 382.15 billion yuan respectively, and it is expected to reach 453.26 billion yuan and 1664.83 billion yuan respectively by 2025, with an average annual compound growth rate of 26.8% and 27.8%, respectively.
The industry is subdivided, and artificial intelligence can be divided into the underlying basic layer, the intermediate technology layer and the upper application layer. Among them, the basic layer mainly includes chips and basic software frameworks, which are an important cornerstone of artificial intelligence; while the technical layer completes the technical support through key technologies and algorithm models; and the final landing is the application layer, which forms software and hardware products or solutions for specific application scenarios.
Glen Deep Hitomi is an AI company that focuses on the technology layer and the application layer. The scale of the computer vision core industry to which it belongs and the scale of related industries have reached 63.33 billion yuan and 143.86 billion yuan in 2019, and are expected to reach 153.71 billion yuan and 485.84 billion yuan in 2025, respectively.
02
The metamorphosis of the "genius company"
As early as the beginning of its establishment, Geling Deep Pupil was regarded as the "genius AI company" in the AI industry, and it was also a fragrant feast in the eyes of investors.
This is not surprising, because Zhao Yong, the founder of Geling Deep Pupil and the core technical personnel of "No. 1", has his own aura. He graduated in computer engineering from Brown University in the United States, worked as a senior researcher at Google Headquarters Research Institute, and was one of the first seven designers of Google Glass.
He also worked as a designer of image processing architectures in the Android operating system.
In 2013 and 2014, Glen Deep Hitomi received a $1 million Series A investment from Zhen Fund and Ceyuan Fund, and a $17 million Series B investment led by Sequoia Capital, respectively.
Later, after the initial stage of core technology formation and commercial exploration, Geling Deep Pupil finally began to take off.
From 2013 to 2022, Geling Deep Pupil completed this unique "growth transformation", from a "genius company" to a trinity AI company that has achieved "far-sighted, stable and upright" from technology, scenario landing to financial data.
Source: Prospectus
03
Seeing Far: Choosing the Hardest Technology Path
From the very beginning, Gren Deep Pupil has been a company that insists on technology first, which is where its "far-sighted" attribute lies.
Externally, the company has open sourced its TrillionPairs and Glint360K face recognition datasets to the academic community during the research and development process, and open sourced the PartialFC training code to promote the development of the industry with its own strength;
Internally, the company has 19 invention patents, 6 utility patents, 4 design patents and 77 software copyright patents.
In the various competitions with benchmarking roles in the industry, Geling Deep Pupil has also achieved good results: the accuracy rate of face recognition is 99.1%, and the 1:1 test pass rate of the 100 million-level portrait algorithm test of the Ministry of Public Security is 99.97% (one in 10,000 error rates).
At present, Geling Deep Pupil has formed five major technical directions, including model training and data production technology based on deep learning, 3D stereoscopic vision technology, automated traffic scene perception and event recognition technology, large-scale cross-border tracking technology and robot perception and control technology.
Among them, the most important difference is that when Companies such as SenseTime and Megvii are still choosing to use modes such as marking and stacking algorithms to let computers learn to recognize faces, Glin Deep Pupil has already chosen 3D vision technology one step ahead of the time.
But this is not without risks, in the early days, the development cost of Grening Deep Pupil was higher, and the time to commercialization was also late. But after looking far, the line is more stable.
The layout of 3D stereo vision technology of Göring Deep Pupil includes multi-eye sensor calibration and depth estimation technology, motion attitude analysis technology, 3D reconstruction and stereo vision analysis technology, of which 3D reconstruction and stereo vision analysis is a key basic technology, and the technological advancement is reflected in point cloud registration, that is, matrix transformation of point cloud data under different perspectives under the same coordinate system, so as to achieve the complete 3D model reconstruction of the target.
In this field, the company has proposed an overlapping point prediction module based on global information and a non-overlapping point filtering module based on Transformer, which effectively improves the accuracy of local and local registration of point clouds.
The data shows that the evaluation index of the rotation angle error (that is, the angle between the reference fixed point, the fitted data and the real image) on the ModelNet40 dataset can reach 1.16°, which is the best level of technical indicators in this field, and the accuracy of the second place is 30%.
Source: Company Announcement "Issuers and Sponsors Reply to Opinions (2021 Semi-Annual Financial Data Update)
Note: Among them, Error and MAE both represent errors, and the smaller the value, the better.
Therefore, it has also become the only AI company in China that can use 3D stereoscopic vision technology to form a downstream commercial landing.
04
Steady: Diversification of landing scenarios
After all, the choice of longer-term technology is still only "looking good", and eventually it will fall to the level of reality, and the key step is still - commercialization.
This is also a problem that plagues almost all AI companies in the entire field of computer vision.
The previous "Four Little Dragons" chose different landing paths for themselves: Megvii Technology chose the Internet of Things, SenseTime chose to do the platform, Yuncong chose finance, and Yitu chose the most traditional security, and later directly turned to the basic layer of AI chips.
For Glenn Deep Pupil, who has chosen a "technical route that few people take", its commercial choice has also become crucial.
According to the prospectus, the downstream covered by the current Deep Pupil of Gehring mainly includes urban management, smart finance, commercial retail, rail transit and sports and health, of which urban management accounts for the largest proportion, but the proportion of income is becoming more and more balanced with the passage of time.
As mentioned earlier, only the "financial" field has cloud from the technology of this competitor, and the other two areas selected by Gren deep pupils can hardly see large competitors.
When the eyes turn to the financial field, there are also significant differences between The Deep Pupil of Glen and the Cloud from Technology:
The cloud mainly relies on the human-machine collaboration system to do the work of front-end customer verification and identity authentication, while the Glen Deep Pupil mainly relies on the three-dimensional behavior recognition and analysis system, which is applied to the security operation scenarios of the vault, the banknote press and the outlet, etc., to carry out compliance monitoring and behavior early warning.
To put it simply, from the early technology choice to the later commercial exploration, Gren Deep Pupil has been seeking differentiation from competitors and seeking to take a more resolute and stable path.
Now, Gerin Deep Pupil has reached a height that is difficult for competitors to reach on its own differentiation path.
05
Litezheng: The financial data is the most robust
Long-term investment in technology and the choice of landing scenarios ultimately need to be reflected in financial data.
Although from the overall report, the current domestic AI industry companies have not been able to achieve large-scale profits, the net profit attributable to the mother in 2018-2020 is also in a state of loss.
However, taking the 2020 data as an example, the net profit attributable to the mother in that year is that if the impact of the share payment fee is excluded, then its non-attributable net profit in 2020 is 10.3169 million yuan, which has achieved a turnaround and is the first AI company in the industry that has achieved operating profits.
If the profitability of each family is still on the road, the difference between per capita income and gross profit margin is already very significant.
From 2018 to 2020, the per capita income generation of Gling Deep Pupil increased from 207,900 yuan per capita to 922,900 yuan, ranking first among comparable companies in the industry.
From 2018 to 2020, the overall average gross profit margin of the industry was 50.06%, 48.21% and 38.71%, respectively, while the gross profit margin of Glading Deep Pupil was 62.60%, 53.13% and 61.57%, respectively, which was much higher than the industry average.
In 2021H1, the company's gross profit margin was as high as 72.37%.
Behind each of these excellent data, it is the choice and accumulation of Gren Deep Pupil for up to 9 years.
End
There will be companies in every industry that are willing to do difficult things, willing to pay higher costs, and wait for the day of "accumulation and thin hair".
In the field of new energy vehicles, BYD has long adhered to lithium iron phosphate and improved its own industrial chain in a vertical mode; in the field of AI, this company may be The deep pupil of Gling, not to do rough model accumulation, willing to build a technical system from scratch.
"Doing the difficult but the right thing" is the sense of mission that every participant in the Fourth Industrial Revolution should have, and Grenkin's deep pupil has already told us the answer with practice.
*The above content is a third-party product announcement, does not constitute investment advice, and does not represent the views of the publishing platform. Users should consider whether any opinions, opinions or conclusions contained herein are consistent with their particular investment objectives, financial situation or needs. The market is risky, investment needs to be cautious, please judge and make decisions independently.