Who will be the next company to go public in the AI field?
Since the second half of 2023, 25 artificial intelligence-related enterprises have gone public through IPO on the Hong Kong Stock Exchange. In the first half of 2025 alone, 5 AI enterprises successfully went public. It can be seen that the AI sector is becoming the most certain growth point in the capital chain.
Capital never acts aimlessly. What commonalities do those “AI favorites” selected by capital possess? Which implementation directions have the most realistic potential for monetization? And how should companies aiming to become the “first AI stock” proceed?
In 2025, AI seems to have become one of the few directions that can still combine certainty and imagination, attracting the accelerated convergence of capital, talent, and narratives.
Data from IT Juzi shows that as of the third quarter of this year, a total of 764 AI companies have received venture capital investment, reaching a new high in the past five years, with the transaction amount reaching as high as 83 billion yuan.
On the surface, the enthusiasm in the primary market is picking up. However, in fact, there are not many investable tracks in the market, and capital still concentrates on the AI field, presenting a unique scenery.
This phenomenon of “concentrated enthusiasm” does not only occur in the primary market. The IPO trends in the secondary market also verify the capital attractiveness of AI. According to the big data statistics of LiveReport, since the second half of 2023, 25 artificial intelligence-related enterprises have gone public through IPO on the Hong Kong Stock Exchange. In the first half of 2025 alone, 5 AI enterprises successfully went public. It can be seen that the AI sector is becoming the most certain growth point in the capital chain.
Capital never acts aimlessly. What commonalities do those “AI favorites” selected by capital possess? Which implementation directions have the most realistic potential for monetization? And how should companies aiming to become the “first AI stock” proceed?
These are precisely the key propositions in the next stage of competition in the AI industry.
The “AI Favorites” Selected by Capital
Compared with 2024, the AI enterprises selected by capital this year are obviously more pragmatic.
According to the statistics of Industrialist, in the first three quarters of 2025, a total of 548 transaction events occurred in the primary market of the artificial intelligence field, a year-on-year increase of 44.59%. Specifically in the segmented tracks, general AI applications and AI industry applications accounted for nearly half, becoming the two most rapidly growing main lines. In terms of transaction amount, AI industry applications jumped to the track with the most concentrated capital, reaching 15 billion yuan.
This trend is completely different from the frenzy of capital in the AIGC layer last year.
In the same period of 2024, the transaction amount in the AIGC track reached as high as 30.8 billion yuan, accounting for 60% of the total. This year, this figure has sharply decreased to 11.1 billion yuan, a year-on-year decrease of 64.01%. The “cooling down” of capital towards AIGC has also directly pulled down the entire AI transaction amount, making the transaction amount in the first three quarters of 2025 only 46.064 billion yuan, a year-on-year decrease of 8.43%.
However, capital has not left the market but has flowed more precisely into the “implementable” fields.
Capital has invested more resources in three major directions: general AI applications, AI industry applications, and AI basic technologies, with increases of 216%, 36.8%, and 153.74% respectively. It is worth noting that although the basic layer has received more financing, this is partly related to the heavy equipment and high investment in this field. Capital is not shrinking but adjusting its allocation, putting money in places where it can be better transformed into productivity.
Further analysis of the enterprises that received more than 100 million yuan in financing in the primary market in the first three quarters reveals that most companies mainly focus on specific application scenarios such as healthcare, logistics, autonomous driving, robotics, and digital marketing. Capital is obviously returning to a principle, that is, AI should be “down-to-earth”.
Now let's look at the secondary market in the AI field.
As of July 2025, five enterprises have successfully gone public in the secondary market, covering multiple fields such as warehousing and logistics, solutions, and unmanned mining trucks.
On the surface, these enterprises also belong to industry applications. However, the difference is that IT Juzi classifies these enterprises into the AI basic layer and the AI technology layer. In addition, these enterprises generally have stable profitability. For example, Geek+ has achieved an income of 1.025 billion yuan, a year-on-year increase of 31.0%, which benefits from the successful implementation of the AI business model in warehousing robots. Although the overall profit growth of Unisound has slowed down, the income of its Shanhai large model is nearly 100 million yuan, a year-on-year surge of 457%. This shows the amplifying effect of AI technology in the original business system.
It is worth noting that whether in the primary market or the secondary market, most of these “AI favorites” do not have a “pure” AI attribute.
In other words, AI is not their only label, and most of them are not AI-native companies. Minglue Technology still focuses on government and enterprise digitization, knowledge graphs, and intelligent decision-making. Unisound was established in 2012, and its positioning has also changed. The foundation of DripTech lies in data analysis. In the primary market, there are very few truly “pure AI” companies. Among the few that exist, all of them have the support of giants or the resources of mature industrial chains. For example, betteryear is backed by Alibaba, and Intent AI has received strategic investment.
Similarly, the winners in the secondary market are often those enterprises that have been deeply involved in the industry for a long time and then achieved a leap with the help of AI.
This reveals an intriguing phenomenon. Whether in the primary market or the secondary market, capital is not betting on AI but on “companies that can be AI-ized”. Companies that truly go public through IPO have clear commercialization paths.
What is the Real Temperature of AI in Industrial Implementation?
The choice of capital is never an isolated event. What it reflects behind is the collective judgment of the market on the difficulty of implementation.
Looking back at the past few years when the AI technology boom arrived, the most and fastest implementation of technology has been in the transformation of the interface layer. This layer of transformation mainly occurs in scenarios such as programming development, customer service, office work, and content marketing. They have the characteristics of “low risk and high frequency”. AI has improved operational efficiency and reduced labor costs in these links, and has also quickly allowed enterprises to initially see positive feedback on ROI.
For example, in the programming development scenario, a study by GitHub shows that AI programming assistants can increase the speed at which developers complete tasks by about 26%. AI is not only good at code completion, unit test generation, and logic optimization but also participates in preliminary code reviews.
However, as AI enters higher-level development links, its shortcomings gradually emerge. An experiment by MIT points out that for senior developers, the improvement of productivity by AI is only 8% - 13%. In some scenarios, the guidance of AI has even extended the development time by 19%.
This also explains why in the primary market, many AI startups focusing on interface layer applications seem to be very popular in the early stage but have difficulty in continuously expanding their revenue curves. Low technical thresholds, strong service substitutability, and serious homogenization are the systematic problems in this track.
Only when these AI transformations at the interface layer are deeply integrated with business processes can they release real productivity. Some enterprises have realized this and started to embed AI into the core business chain. As a result, in fields such as healthcare, finance, education, logistics, and retail, the focus of AI is gradually shifting to the transformation of the process layer.
The healthcare industry is a typical example. The four-quadrant model proposed in the Alibaba Cloud white paper shows that medical imaging diagnosis and drug R & D are in the “high maturity + high potential” range. Among them, the commercialization rate of medical imaging AI products has reached 90%. Insilico Medicine's generative AI platform has compressed the antibody R & D cycle by 10 times and achieved four drug licensing cooperation agreements, with a total amount of more than $1.5 billion. And this enterprise has received multiple large-scale financings this year.
The implementation of AI in the healthcare industry is not limited to R & D. BD has optimized inventory management through predictive analysis, increasing the prediction accuracy by 20% and significantly reducing inventory costs. The intelligent hospital solution of Senyi Intelligence has shortened the doctor's documentation time by 50% through AI-driven medical record generation and quality control. Siemens' AI platform has improved the diagnostic accuracy and increased the efficiency of the radiology workflow by 40%, achieving collaborative optimization.
The common point behind these cases is that AI is no longer just a tool but a decision-making variable embedded in the system. Once it enters the process, it becomes the deepest moat for enterprises and is therefore the direction that capital is most willing to bet on.
However, the transformation of the process layer is not easy. Firstly, in terms of data, enterprise data is scattered in different systems with inconsistent formats. AI models need a large amount of cleaning and annotation to understand it. Secondly, in terms of organization, allowing AI to take over the decision-making process means that management needs to delegate power to algorithms. Most enterprises are still in a “dual-track” state of “AI + human”. Model decisions still need human verification.
The complexity is highlighted in the healthcare scenario. Problems such as data bias, insufficient interpretability, unclear ethical responsibilities, and complex process embedding have made AI mainly play roles such as quality control, image interpretation, and doctor-patient communication in the healthcare industry at present, and there is still a significant distance from “independent diagnosis”.
However, from the perspective of the technological rhythm, this is not the end of AI implementation.
The real value of AI lies in the redefinition of business logic. This layer is not about tool upgrading or process optimization but about the reshaping of the entire system operation mechanism. Autonomous driving, embodied intelligence, and AI shopping guide systems built by JD.com and Kuaishou, as well as the “AI partner” of Alibaba.com, are essentially exploring an AI-native organization and operation mode.
But this step is also the farthest. It requires not only high-quality data and huge computing power investment but also years of engineering implementation verification. At present, only a few companies can reach this layer, and most are still in the laboratory verification or partial pilot stage.
Generally speaking, it is difficult for AI at the interface layer to continuously realize monetization. Investing in AI at the business logic reshaping layer is for the future, while AI at the process transformation layer may be the most easily realizable commercialization range at present.
This also explains why there are very few truly “pure AI” companies in the primary market, and most listed enterprises classified as “AI concept stocks” in the secondary market are actually a combination of “AI + years of industry experience”.
Searching for the Next IPO in the AI Wave
In the AI field, the next company to ring the bell is accelerating its arrival.
Data shows that as of July 24, 2025, 214 companies have submitted listing applications to the Hong Kong Stock Exchange, among which as many as 48 are AI-related enterprises, accounting for 22.43%. In other words, one out of every five companies planning to go public is related to AI.
What's more interesting is that among these 48 AI concept companies planning to go public, 11 are A + H companies, accounting for as high as 22.92%. This means that a group of “leading enterprises” with mature technology, clear commercialization paths, and the ability to conduct international layout have emerged in the AI field. They are strengthening their capital foundation and expanding their global influence through dual listing.
There are numerous heroes in the world. Who can stand out and become the next real industry bell ringer?
Looking back at a group of AI companies that successfully entered the capital market around 2025, an implicit consensus can be found: the implementation path should be “from point to area”, and the technical approach should be “starting from scenarios”.
Whether it is Unisound's healthcare AI, Minglue's marketing AI, Geek+'s warehousing robots, or Sian Intelligent Driving's port unmanned vehicles, they almost all follow a similar trajectory, that is, starting from a business scenario with high certainty, deeply exploring needs, polishing products, achieving leadership in the industry and then gradually expanding horizontally to similar fields.
Unisound chose medical record quality control and subway voice ticketing as its breakthrough scenarios. By establishing benchmark projects, it has built an early user perception of usability and goodness, providing a practical starting point for the subsequent application expansion of general large models. Haizhi Technology also follows a similar path. It entered the financial anti-fraud scenario through knowledge graph technology and gradually extended its capabilities to operational risk control and data governance, achieving a virtuous cycle of “scenario breakthrough - platform precipitation”.
In the choice of commercialization paths, these AI enterprises generally avoid direct competition with general large models and instead adopt a “vertical + specialized” strategy.
It is worth noting that this group of AI enterprises at the forefront of “ringing the bell” also have a commonality, that is, they almost all invested in building their own platform products or basic technical architectures at an early stage.
For example, Unisound has its self-developed large model platform “Unisound Brain” and voice chip Atlas. DripTech has built the FastData and FastAGI dual platforms, corresponding to the data engine and intelligent engine respectively. Minglue has built an end-to-end AI operation system from analysis to decision-making between the “Miaozhen System” and “Xiaoming Assistant”. Geek+ and Sian Intelligent Driving have built a full-stack solution integrating software and hardware around AI application scenarios, achieving an end-to-end closed-loop from algorithm, perception, decision-making, to execution.
On the one hand, these self-owned platforms condense core algorithms, models, and toolchains, forming technical barriers. On the other hand, they also facilitate the standardized replication of products and reduce project delivery costs. It is a necessary stage for AI enterprises to develop from labor-intensive projects to productization and scale. The stronger the platform ability of an enterprise, the more it can dominate the commercialization rhythm.
According to the predictions of institutions such as IDC, the Chinese AI solutions market will still maintain an annual growth rate of over 50% in the next five years, and its scale is expected to exceed one trillion yuan in 2030. This is both a huge opportunity and a fierce battlefield.
In the era of large models, technology is changing rapidly. Only by mastering unique scenario data and know-how can enterprises resist homogeneous competition and become the next company to ring the bell.