The Heart of the Machine is original
Author: Wu Xin
At the business level, compared with peer suppliers, the differentiation and characteristics of Innovation Qizhi in the AI+ industry are reflected in the "brain, eyes, hands and feet". The industry track and business model determine that it is different from the "weight" of traditional AI companies based on security, which is significantly reflected in financial data.
On January 3, Qingdao Innovation Qizhi Technology Group Co., Ltd. (hereinafter referred to as "Innovative Qizhi") once again applied to the Hong Kong Stock Exchange for listing, updated the prospectus, and opened a public offering this week. According to market news, the company has been fully subscribed for an international placement.
In February 2018, Innovation Qizhi was officially established, with Kai-Fu Lee as chairman. Serving as chairman of the AI company in which it invests is the first and only time so far.
Unlike the "AI Tigers", which were established earlier and focused on "black technology" and security applications, Innovation Qizhi began with enterprise-level applications, mainly for manufacturing and financial services. The head of the company is also not from a research and development background, but an entrepreneur with more than 20 years of experience in corporate services.
From its inception to becoming a unicorn, it took only three years to innovate Qizhi. In less than four years, the company's commercial revenue scale has ranked at the forefront of artificial intelligence startups.
According to Sullivan data, in terms of revenue in 2020, innovation Qizhi is already the largest SUPPLIER of AI-driven solutions in China's manufacturing AI solution market. The 3rd largest AI technology-driven solution provider in China's enterprise AI solution market.
This time to create a new customs, innovation Qizhi is expected to become the first "AI + manufacturing" stock landed on the Hong Kong Stock Exchange.
First, read the fundamentals of data
Since its inception in 2018, as of January 3, 2022, the last round of financing before the listing, Innovation Qizhi has conducted a total of six rounds of financing. The angel round raised over 100 million yuan, and the last round of D financing was invested by SoftBank Vision Fund Phase 2.
Kai-Fu Lee serves as chairman of Innovation Qizhi and does not directly hold the company's shares. Innovation Factory holds a combined stake of about 30.01%, the single largest shareholder, and other major shareholders include CICC 16.7% and SoftBank holding 7.12%.
The company's main revenue comes from product and solution sales, data solution services.
The data shows that in 2018, 2019, 2020 and the first nine months of 2021, the operating income was 0.37 billion, 229 million, 462 million and 553 million yuan, respectively.
Sorted out according to the company's prospectus. Data for 2021 is current to September 30.
At the same time as revenue growth, innovation Qizhi inevitably fell into losses.
Net adjusted losses, net adjusted losses were gradually narrowed to $0.45 billion, $160 million, $144 million and $81 million, respectively, excluding non-operating items such as share payment expenses and financial expenses arising from callable share financial liabilities. The adjusted net loss ratio has dropped significantly from 121.6% in the early days to 14.6%, and from a trend point of view, the loss is being quickly offset by the cost reduction brought about by the growth of business income and the scale effect, and the ability to reduce losses is performing well.
The data shows that the loss is mainly due to the continuous expansion of the company's business scale, and the general administrative expenses and research and development expenses have also increased. Among them, Innovation Qizhi invests a lot of resources in the research and development of AI technologies, including computer vision and machine learning.
For example, R&D expenditure was 0.29 billion in 2018 and 177 million in the first nine months of 2021. In the past fiscal year, R&D expenditure still constituted the company's largest expenditure, and the primary purpose of the funds to be raised in this IPO is also research and development.
According to the prospectus, about 255 employees of the company are engaged in artificial intelligence and technological intelligence, accounting for 69.1%. At the technical level, R&D efforts are focused on areas with potential commercialization opportunities, fewer barriers to competition and cost savings, so as to optimize the company's resources optimally. For example, smart molten iron transportation and defect detection solutions have been recognized by manufacturing customers and end users.
Continuous investment in research and development has also led to a wealth of intellectual property rights. The company has applied for 634 AI-related patents and successfully registered 126 AI-related patents, including 79 invention patents. It is at the leading level among its peers in the industry.
Although it is not profitable, the company's administrative expenses and research and development costs are narrowing. As a company that has only been established for more than 3 years, its operating efficiency is already at an excellent level in the industry.
During the reporting period, the company's gross profit increased steadily, to 0.23 billion, 0.72 billion, 134 million and 170 million, respectively. Gross margin has stabilized at around 30% since 2019.
Among them, September 30, 2021 increased by more than 2% compared with the same period on September 30, 2020, an increase of nearly 1.8 percentage points from the gross margin for the whole year of 2020.
The company's manufacturing revenue has become the main force of the company's revenue, but also because of the soft and hard business model, it will naturally affect the gross profit level.
For the change in gross margin, Innovation Qizhi explained in the prospectus that:
This is mainly because most of the AI-based products and solutions sold in 2018 are software-based solutions. Since 2019, sales have focused on software and hardware integration solutions with more hardware components, with the latter typically having lower gross margins than the former.
Looking at the change in cost of sales, material cost spending since 2019 has indeed increased significantly, as more hardware purchases are involved. The cost of purchasing hardware is high, which naturally increases the operating cost.
In fact, gross profit margins are usually determined by the industry track and business model in which the company is located, some companies choose the saas model, some companies take the platform road, and some focus on the solution market.
For example, SenseTime positions itself as an artificial intelligence software platform enterprise, while Innovation Qizhi is positioned as an artificial intelligence product and solution provider, and its focus on manufacturing belongs to the typical "heavy scene" model.
Generally speaking, all AI solution products are a set of software, hardware and technical services in one, but the scene has "light" and "heavy", relative to the typical "light scene" such as security, new retail and automatic driving, industrial manufacturing belongs to the "heavy scene" of AI applications, the heavier the scene, the higher the proportion of hardware in the final AI solution, which will directly determine the gross profit margin level.
At the same time, the business model adopted by the enterprise is closely related to the maturity of the industry.
Early in the application of new technologies, tailor-made solutions are often needed to make the technology relevant and lead to initial adoption. For AI, customized solutions are still important for establishing application relevance, but the key to scalability is to standardize the solution into a modular product.
In particular, the demand for China's manufacturing market is highly fragmented and non-standardized. Selling only technical solutions, as in the late 19th century as selling conveyor belts to blacksmiths, simply did not work. In anticipation of the Industrial Revolution, a factory was built around a conveyor belt.
Hardware and software are just as important to companies that are on the move. Usually the first thing to do is lay out the hardware infrastructure. The standardization of the hardware level will bring about the standardization of the later software level. If the hardware is not standardized, the standardization of the software is empty talk.
The "soft and hardware integration" solution can usually be flexibly integrated into the original production line of the enterprise, without special transformation of the production line, which can reduce the difficulty and threshold of the application of AI in the enterprise, and can also complete the project delivery faster.
This is also why in the past few years, whether it is an Internet manufacturer or a start-up company involved in the vertical field, almost all of them are accompanied by "soft and hard integration", Baidu, Tencent and other Internet manufacturers have also launched industrial quality inspection equipment. Innovation Chichi also built its own moat in this way.
In the analysis of corporate health, in addition to revenue, gross profit, etc., important data also includes comprehensive cash flow performance.
When a business doesn't have enough book money to last a year, you don't know how panicked it is. Innovation Qizhi does not have such concerns, and the company's cash flow is relatively stable.
As of September 30, 2021, 16. A total cash balance of 500 million is sufficient to cover its operating activities and provide sufficient liquidity for the expansion of business operations.
Performance in terms of consolidated cash flow.
In recent years, the company's inventory and inventory amount have increased significantly. During the Reporting Period, the inventory amount of Innovation Qizhi was 3.043 million yuan, 32.327 million yuan, 55.31 million yuan and 43.359 million yuan, respectively.
The company's inventory is dominated by work in process and raw materials, which is a symbol of the opening of the channel market, indicating that the company's AI products and solutions have increased sales, and new project stockpiling has led to an increase in inventory.
Inventory turnover days were 36 days, 41 days, 49 days and 35 days. Compared with some AI companies focusing on 2G business, the inventory turnover days due to strategic stockpiling due to Sino-US trade frictions have been lengthened, and the inventory turnover days of Innovation Qizhi are at a lower level.
The opening of the business situation was also accompanied by an increase in accounts receivable. During the reporting period, the accounts receivable of Innovation Qizhi were 16.24 million, 120 million, 190 million and 296 million, respectively. This is related to the characteristics of the industry itself, and manufacturing customers generally need longer settlement periods than other industries.
The number of receivables turnaround days also increased accordingly, to 76 days, 115 days, 135 days, and 137 days, respectively. Compared with the accounts receivable cycle of AI companies with 2G business-based AI companies with more than 290 days and more than 130 days, the number of days of innovation qizhi is relatively low, and it is at a low level among AI companies.
In addition, the majority of the Company's accounts receivable are less than 6 months old, and as of September 30, 2021, approximately 85.3% of trade receivables are less than 6 months old.
Second, the business experience of "AI+ manufacturing"
In the case of difficult commercialization of new technologies, the choice of industry and scenario is very important.
The so-called scene ability is not entirely a sales problem, but also means how to find a certain segment forward and have the opportunity to form a monopoly or a half.
This is also why when machine vision AI companies frantically gathered in the fields of security and retail, Innovative Qizhi became the first batch of people to eat crabs in "AI + Manufacturing".
At present, manufacturing is a field with a huge market volume and few players. However, if China's manufacturing industry wants to continue to develop, intelligent manufacturing that integrates information and communication technology with advanced manufacturing technology is the only way. Improving production efficiency and moving towards high-end manufacturing is also the road that manufacturing powers such as Japan and Germany have taken.
At the end of 2021, the Ministry of Industry and Information Technology and other eight departments jointly issued the "14th Five-Year Plan" Intelligent Manufacturing Development Plan, which pointed out that by 2025, most of the manufacturing enterprises above designated size will achieve digital networking, and the backbone enterprises in key industries will initially apply intelligence.
The plan also pointed out that to enhance the supply capacity, the market satisfaction rate of intelligent manufacturing equipment and industrial software exceeded 70% and 50% respectively, and more than 150 system solution providers with high professional level and strong service capabilities were cultivated.
Artificial intelligence solutions are beneficial and infiltrated in the aspects of "product quality control consistency, process management efficiency, labor costs, safety control, and integrated management capabilities" of manufacturing enterprises.
The proportion of revenue generated by MANUFACTURING AI products and solutions is increasing, accounting for a large proportion of total revenue.
In the core track of "AI + Manufacturing", the company is deeply engaged in panel semiconductors, iron and steel metallurgy, energy and power, engineering and construction, automotive assembly, high-tech / 3C and other vertical segments, mainly serving the transformation needs of operational efficiency and information intelligence.
Innovative Smart Manufacturing System (AIMS)
In the core track of "AI + Finance", the company is not engaged in front-end services such as risk control and marketing, but related to the IT infrastructure of insurance and banking, such as intelligent data center construction and operation and maintenance, intelligent hybrid cloud management, etc.
In addition to being very different from the traditional "AI Four Dragons", compared with other suppliers in the same industry, Innovation Qizhi is one of the few suppliers of full-stack artificial intelligence products and solutions that integrate "AI Industrial Vision - AI Industrial Automation - AI Industrial Cloud Platform".
At the product level, they not only built the machine vision intelligence platform ManuVision, the edge video intelligence platform MatrixVision, but also built the Orion distributed machine learning platform, and are one of the few companies in China with a proprietary deep learning platform.
The relationship between three proprietary AI platforms.
"Our differentiation and specialization in the AI+ industry is reflected in the 'brain, eyes, hands, feet'." Xu Hui once explained.
Orion's distributed machine learning platform is like a "brain"; AI industrial vision is like an "eye", responsible for locating, identifying, measuring and detecting. Moreover, these platforms are not pure software platforms in the traditional sense, but can be combined with the "hands" and "feet" of robots.
For example, ManuVision can be compatible with industrial light sources, industrial cameras, manipulators, controllers and other external hardware equipment, for customers to create a set of optical (learning), machinery (machinery), electricity (gas), software (parts), arithmetic (method) in one of the automation overall solutions.
Innovate how Chichi integrates LEGO bricks into some AI products and solutions.
Contrary to the "top-down" path of early AI companies that "platform first, then empower", Innovation Qizhi did the opposite from the beginning - taking "bottom-up", that is, "solutions-products-platforms".
The company started from the specific application scenario, through the cooperation with the industry representative large and medium-sized enterprises benchmarking cases, and gradually modularized the replicable elements and the industry know-how, thereby achieving extended replicability, while avoiding the path of system integrators.
Although Innovation Qizhi also has an AI platform, unlike the platform-based thinking of early AI companies, Innovative Qizhi's AI platform is mainly used as the underlying AI infrastructure for the company to serve customers, as a sedimentation pool for AI technology assets.
For example, the algorithm model for defect detection currently developed for the ManuVision platform has increased from 215 classes to 282 classes at the end of June 2021, and the model for field inference algorithms developed for the MatrixVision platform has increased from 277 classes to 352 classes.
These technology assets are highly condensed, lowly coupled and highly reusable to support and enable efficient project delivery that meets customization requirements. Thanks to these technology assets, some aspects of the workflow can also be automated.
In addition, some technology solutions can be reused across domains to similar application scenarios.
Innovation Qizhi is not positioned to be a single product supplier or divergent system integrator, but to deepen the industry, the combination of technology focus and industry needs, the formation of a series of product solutions to meet the needs of the general industry.
In practice, they usually transplant the "brain" trained model to the corresponding image or video AI platform, and then embed it in the company's ABS and RDP to realize the customer's needs on the spot.
The so-called RDP (Rapid Deployment Product), the degree of productization is relatively high, can be quickly deployed. ABS (Assets-based Solution) is a solution that is suitable for combining with the customer's industry scenario, and this solution can also be deployed relatively quickly.
However, ABS or RDP products are usually relatively low in terms of unit price, what to do for large customers and head customers?
For example, if you want to go to a large household to do business, you must not be satisfied with sending milk, you must have done the laundry, the lawn has been cut, and finally made into a customized and long-lasting, connected butler service, which is the Chinese version of IBM.
Innovative Qizhi is to lock in large customers by leveraging these reusable, asset-based RDP/ABS portfolio differentiated industry solutions to provide long-term service to customers' various scenario needs.
From this, we can also see the strategy of innovating the digital transformation of the whole life cycle of qizhi services customers, and the clear path to improve the overall profitability level in the future.
It is worth noting that in addition to being satisfied with its own use, the company is also building an ecosystem through three major platforms.
The prospectus reveals that the company is building an open architecture technology platform to attract more players in the AI industry value chain and deepen cooperation with various industry players.
The ManuVision and MatrixVision platforms, currently embedded in ABS and RDP deployed at the edge of the customer's site, will evolve into cloud-based and more compatible platforms to connect more devices and applications of different industry players over the wireless or 5G network, and support compatibility with third-party plug-ins to attract more players in the AI industry value chain.
In addition, in order to deepen the capabilities of the existing customer base, they have also innovated in the cooperation model - establishing joint ventures with leading enterprises in the industry to expand the results of cooperation to more scenarios.
CCID Qizhi, a joint venture with MCC CCID Group, is a success story. "The strategic partnership not only allows us to gain insights into its industry, but also allows us to leverage its industry resources to quickly penetrate into relevant industry verticals." The prospectus explains.
The increase in innovative Qizhi high-quality customers is obvious.
This model has also driven Innovation Qizhi to quickly acquire a large number of users in a short period of time, and the total number of company customers has increased from 50 in 2018 to 157 in 2020.
The share of manufacturing in total revenue has also continued to increase, from 36.6% at the end of 2018 to 50.8% in the first three quarters of 2021.
As of December 31, 2020, Innovation Qizhi had 23 high-quality customers, accounting for 82.5% of the contribution, and the repurchase rate of high-quality customers in terms of amount was 112.7%.
3. Risks and potential
The ability of Innovation Qizhi to become the first "quasi- " listed "AI + manufacturing" company on the Hong Kong Stock Exchange is already a comprehensive affirmation of its company's valuation, profitability and development prospects.
According to Sullivan data, in terms of revenue in 2020, Innovation Qizhi is the third largest AI technology-driven solution provider in China's enterprise AI solution market and the largest AI technology-driven solution provider in China's manufacturing AI solution market.
According to IDC data, Innovation Qizhi is the fourth largest machine learning platform vendor in China from 2020 to 2021H1 and the fifth largest in china's computer vision application market share in 2021H1.
It is worth noting the risk of industry competition. On the one hand, the enterprise AI solutions industry is highly competitive, with competitors typically including:
1) Compete with other companies focused on developing and commercializing AI technologies, 2) compete with existing players who are not focused on AI in the various verticals in which they are already involved, 3) compete with new industry entrants, and 4) potential competition from global tech companies.
On the other hand, the market is very fragmented. In terms of 2020 revenue, while Innovation Qizhi has become the third largest technology-driven supplier of AI solutions for enterprises in China, it has a market share of only 0.3% of the more than 1,500 market players.
However, the cake of the whole market is still getting bigger.
According to Frost & Sullivan, the market size of China's enterprise AI market reached approximately RMB139.4 billion in 2020 and is expected to reach approximately RMB836.6 billion by 2025, representing a COMPOUND annual growth rate of 43.1%.
In 2020, China's enterprise-level AI market will account for about 75.0% of the entire AI market, and it is expected to increase to about 80.0% by 2025.
According to other data, with the fuller adoption of digital and intelligent technologies, the artificial intelligence solution market for China's manufacturing industry is expected to reach about 64.9 billion yuan by 2025, with a compound annual growth rate of 48.3% from 2020 to 2025.
"The industrial potential of AI mainstream deep learning technology can be 'eaten' for at least 20 years, and it will penetrate into a variety of different fields." Kai-Fu Lee was optimistic that AI would be completely popularized and enter every company.
In the process of AI in manufacturing, finance and other industries, who can take the lead in listing, get continuous support in the capital market, and who can seize more AI market dividends.