Almost doubling the IPO price, is SenseTime Group Inc. a good company to invest in?
SenseTime Group Inc. (“SenseTime”), a leading AI technology company, launched its IPO on 30 December 2021 in Hong Kong. The IPO price was estimated from HK$3.85 to HK$3.99. After the IPO, the share price increased to HK$8.20 on 4 January 2021 and maintained HK$7.17 on 10 January 2021.
Such a big increase in the share price shows the positive attitude from investors and appreciation from the market, despite that SenseTime is still in a loss position of RMB3.7 trillion (USD583.7 million) for the first-half year of 2021. However, will this trend be sustainable in the future and is SenseTime really a target to invest in?
AI Industry overview
Let’s look at the general picture of the AI industry first.
The AI industry ecosystem is based on three layers, the foundation level, the technical level and the application level. The foundation level refers to the chips and database as physical or intangible inputs. The technical level is to make computers understand requests and process data. Finally, at the application level, people apply mature technologies to different scenarios or industries, with computers thinking like humans and solving problems.
AI software markets can be classified as computer vision, speech recognition and natural language processing, and data science, according to Frost & Sullivan. Computer vision enables computers to analyze digital images or videos in order to extract data, perform analysis and automate certain tasks. Based on the prospectus of SenseTime IPO released on 4 December 2021 (“Prospectus”), computer vision accounts for 57% of the total AI software market in 2020 in China and is expected to increase to 61% in 2025.
AI technologies seem to have the ability to penetrate to every industry. On the other hand, lacking concentration also shows that the economy of scales may not be easy to establish, which causes breakeven or payback issues for most AI companies.
In addition, current AI technologies are all based on the training process and require massive power and data. In essence, computers are self-learning from the past. That also reveals the problem to predict the future.
SenseTime: Business analysis
Understanding some basics of the AI industry, we can dive into the business analysis of SenseTime.
SenseTime is proud of its academic excellence, which is a useful tool at the technical level of the AI industry ecosystem. As of June 30, 2021, 40 professors led their research efforts, and approximately two-thirds of over 5,000 employees were scientists and engineers.
Indeed, SenseTime has a powerful team to improve its AI technologies. However, from labs to customers will be another story. Considering applications of AI technologies and the development of AI markets are still far from mature, SenseTime will need more time to enable its intelligent employees to empower its growth.
In terms of the market share, SenseTime did impress us with its outstanding results, with about 15% of the market share in the computer vision software market in 2020 in China.
According to the Prospectus, SenseTime is a leading AI software company with a focus on computer vision technologies, serving a broad range of industries. In other words, they do not have industry focuses yet. Based on a comment from SenseTime’s employee on Zhihu, a Chinese social question-and-answer platform, projects were not easy to complete and most of them are killed in the end. Even if it looks nice to cover as many industries as possible, SenseTime may not have the ability to achieve economies of scale. After all, every industry has its unique features.
The percentage of revenue attributable to its five largest customers increased to 59.3% for the six months ended June 30, 2021 (“1H2021”) from 28.4% for the year end of 2018 (“FY2018”). The customer concentration may show that SenseTime does not have a successful expansion story yet.
Although SenseTime is in the leading position now, it does not mean that it won’t be defeated by newcomers. The current AI technology is based on the training process of massive data, meaning the past. However, there is no data for training to analyze long-tail problems, which is common for many industries. If a new model can solve this issue, it will be a disruptive event for existing players, including SenseTime.
SenseTime: Revenue analysis
Moving from the business analysis, SenseTime’s revenue lines, especially the Smart City segment, have some interesting points to be discussed.
SenseTime has four revenue lines as follows,
1) Smart Business: for enterprises to build AI-integrated workflows and operations.
2) Smart City: for governments to monitor the conditions of public facilities, detect incidents and track the impact of natural disasters.
3) Smart Life: for IoT device, semiconductor, mobile app and gaming companies to enrich users’ experience.
4) Smart Auto: for automobile companies to develop and enhance autonomous driving capabilities.
The Prospectus showed the revenue breakdown from FY2018 to 1H2021 as follows.
It is interesting that Smart City, governments as end users, accounted for nearly half of the total revenue for 1H2021. The total revenue of Smart business and Smart City, from big companies and governments, attributed to nearly 90% of the total revenue for 1H2021.
The biggest issue about the high revenue contribution from Smart City is that it is not easy to sustain the growth in the future.
In the Prospectus, SenseTime already mentioned that uncertainty and changes in government policies and government spending on AI-related products and services will negatively affect the business. There is no doubt that the increasing number of cameras installed everywhere creates privacy and moral issues, which is part of Smart City as a concept. Gladly, this issue is being aware and discussed by the public and governments are more conscious of implementing more AI-related infrastructure in cities.
As governments are spending less and slower, SenseTime may have difficulties acquiring new government clients and increasing its Smart City segment revenue.
Another issue with revenue from governments is how long SengTime can receive the money from governments.
The Prospectus showed the aging analysis of trade receivables based on the date of revenue recognition as follows.
The trade receivables due for more than two years increased to 10% of the total trade receivables for 1H2021 from 1% in FY2018. The long due trade receivables really worsen the cash flows of SenseTime, which is making huge losses.
Considering current client base with governments and large-scale companies are for customized jobs, SenseTime aims to achieve economy of scales using SenseCore, a proprietary AI infrastructure. According to the Prospectus, SenseCore is built on, 1) large-scale supercomputing power, 2) massive data processing and desensitization; and 2) shared platforms and production tools for developers.
SenseCore was only known to the public in July 2021, and we will see whether it can help SenseTime achieve its mass production strategy in the future.
Conclusion
Going through this whole journey of AI market overview, Sense Time business analysis and revenue analysis, it is not difficult to conclude that SenseTime needs more time to prove its worthiness.
Schedule a DDIChat Session in Financial Markets and Analysis:
Apply to be a DDIChat Expert here.
Work with DDI: https://datadriveninvestor.com/collaborate
Subscribe to DDIntel here.
