---
title: "The real deal in the 'AI disillusionment phase': Just raised 50 million and now heading to Hong Kong! Is Haiqing Zhiyuan really that strong?"
type: "Topics"
locale: "zh-CN"
url: "https://longbridge.com/zh-CN/topics/32697747.md"
description: "After more than two years of explosive growth, the AI industry seems to be cooling down by mid-2025 due to technological bottlenecks, market expectation adjustments, and challenges in application scenarios. Gartner's report indicates that AI has entered the &#34;trough of disillusionment,&#34; with companies shifting from blind enthusiasm to a more pragmatic &#34;human-machine collaboration&#34; model, focusing more on data governance, multi-cloud architecture, and vertical application scenarios. Gartner's latest report points out that global AI has entered the &#34;trough of disillusionment&#34;—capital is no longer blindly chasing concepts, and companies are paying more attention to the practical implementation of &#34;human-machine collaboration&#34;..."
datetime: "2025-08-07T06:41:41.000Z"
locales:
  - [en](https://longbridge.com/en/topics/32697747.md)
  - [zh-CN](https://longbridge.com/zh-CN/topics/32697747.md)
  - [zh-HK](https://longbridge.com/zh-HK/topics/32697747.md)
author: "[港股研究社](https://longbridge.com/zh-CN/profiles/3199113.md)"
---

> 支持的语言: [English](https://longbridge.com/en/topics/32697747.md) | [繁體中文](https://longbridge.com/zh-HK/topics/32697747.md)


# The real deal in the 'AI disillusionment phase': Just raised 50 million and now heading to Hong Kong! Is Haiqing Zhiyuan really that strong?

After more than two years of explosive growth, the AI industry seems to be cooling down by mid-2025 due to technical bottlenecks, market expectation adjustments, and challenges in application scenarios.

The Gartner report points out that AI has entered the 'Trough of Disillusionment,' with companies shifting from blind pursuit to a more pragmatic 'human-machine collaboration' model, focusing more on data governance, multi-cloud architecture, and vertical scenario applications.

The latest Gartner report indicates that global AI has entered the 'Trough of Disillusionment'—capital is no longer blindly chasing concepts, and companies are more focused on the pragmatic implementation of 'human-machine collaboration,' with data governance, multi-cloud architecture, and vertical scenario applications becoming new keywords.

As the industry gradually returns to rationality, recently, China's multispectral AI industry leader Haiching Zhiyuan submitted a main board listing application to the Hong Kong Stock Exchange. It is worth mentioning that on July 18, Haiching Zhiyuan just signed a capital increase agreement with D-round investor Shanghai Zhizhi No. 9 Private Equity Fund, with 50 million financing arriving by the end of the month, indicating that Haiching Zhiyuan's ability to attract capital is still relatively strong.

Public information shows that Haiching Zhiyuan uses its self-developed technology in the fields of multispectral sensing and AI algorithms to provide products and services that can detect physical information visible and invisible to the human eye, including multispectral AI modules, multispectral AI sensing terminals, and multispectral AI large model services.

**Founded in 2013, exploded in 2024, Haiching Zhiyuan's 'slow burn story'**

In the AI industry of 2025, some pioneering companies have already provided a high-scoring performance growth report, with AI technology being the core driving force of growth.

According to the prospectus, Haiching Zhiyuan's revenue in the past three years has fluctuated, but the multiple growth is more prominent: in 2022, revenue was 225 million yuan, in 2023, it temporarily declined to 117 million yuan due to industry adjustments, but in 2024, it rebounded strongly to 523 million yuan, a year-on-year increase of 346.29%; in the first quarter of 2025, its revenue further increased to 82.42 million yuan, a year-on-year increase of 282.8%. Corresponding gross profits were 50.88 million, 14.31 million, and 98.17 million, showing significant growth.

After the scale of income and earnings accelerated, Haiching Zhiyuan's profitability also significantly improved: in 2022 and 2023, the company's net profits were -4.93 million yuan and -18.41 million yuan, respectively, but in 2024, it turned losses into profits, with a net profit of 40.41 million yuan;

Adjusted net profit also turned from -2.68 million yuan in 2022 to a significant profit of 42.94 million yuan in 2024; in the first quarter of 2025, adjusted net profit was 16.26 million yuan, compared to a loss of 230,000 yuan in the same period last year, reinforcing the escape from the shadow of losses.

The 'core engine' of this growth is Haiching Zhiyuan's multispectral AI large model service. The prospectus shows that the gross profit margin of this service is significantly higher than that of traditional businesses, and as its commercial application accelerates, it will continue to be the main driving force for revenue growth in the future. In 2024, the release of revenue from the company's multispectral AI large model service drove the overall gross profit margin from 12.2% in 2023 to 18.8% in 2024.

This explosion is not accidental. Haiching Zhiyuan's technical accumulation and scenario landing capabilities are the key to its ability to traverse industry cycles.

Since its establishment in 2013, Haiching Zhiyuan has always focused on multispectral sensing and AI algorithms. With years of technical accumulation in the multispectral AI industry, it has built a full-link technical system covering 'optical sensing and computing,' that is, through self-developed multispectral sensing and computing integrated technology, lightweight edge AI computing, and multispectral AI large model platform, its products can detect physical information invisible to the human eye and achieve 'high-dimensional perception' in traditional security solutions, as well as in smart cities, data center security, industrial monitoring, and other general security scenarios, with strong applicability of its multispectral AI solutions.

These specific products and services include multispectral AI modules, multispectral AI sensing terminals, and multispectral AI large model services.

Currently, Haiching Zhiyuan has served over 3,000 industry customers, including well-known AI companies, telecom operators, and dozens of listed companies at home and abroad; its technology covers the full spectrum of visible light, near-infrared, and terahertz. As of July 28, 2025, it has obtained 95 invention patents, 43 software copyrights, and participated in drafting about ten national and group standards for the general security industry.

These 'hard strengths' have enabled it to gain customer recognition at the beginning of the industry's upward cycle, driving the rapid scale-up of revenue.

Haiching Zhiyuan was recognized as a national-level specialized and new 'little giant' enterprise in 2022, with major customers including China's leading state-owned telecom operators and leading listed AI companies.

**Multispectral AI surge: the dual game of technical barriers and market positioning**

The rise of Haiching Zhiyuan is inseparable from the high prosperity of the multispectral AI track and its 'positioning advantage' in the industry.

Thanks to the explosion of the global multimodal AI market, multispectral AI is in a golden growth period from 'proof of concept' to 'scale landing.' Gartner predicts that the global multimodal AI market size will grow from $2.4 billion in 2025 to $98.9 billion in 2037; among them, multispectral data processing, which can break through the limitations of human eye perception, will become a core growth point.

The Chinese market is even more promising. The Frost & Sullivan report shows that the size of the Chinese multispectral AI market is expected to soar from 20 billion yuan in 2025 to 58.8 billion yuan in 2029, with a compound annual growth rate of 31.0%.

This growth is driven by two major demands: one is the upgrade of 'general security' demand, where traditional security only covers visible light scenarios, while multispectral technology can achieve 'all-weather, all-element' monitoring (such as night infrared recognition, concealed facility millimeter-wave detection); the other is the deepening application of vertical scenarios, from smart cities, campus management to data center security optimization, industrial equipment monitoring, multispectral AI is transforming from an 'auxiliary tool' to a 'core decision support.'

In terms of competition, the entire market presents a domestic leading and international competitive situation.

In the domestic market, the 'differentiated barriers' of the leading Haiching Zhiyuan are relatively prominent: according to the prospectus and industry data, its market share ranks first in the domestic multispectral AI industry, first in the field of multispectral AI large model services, and third in the multispectral AI module market share.

This leadership stems from the dual advantages of 'technology + scenario': technically, its 'optical sensing and computing' system achieves deep integration of multispectral sensing and AI algorithms, solving the pain points of 'low perception accuracy and high computing cost' in traditional solutions; in terms of scenarios, the company has deeply cultivated the general security field, accumulated real data from over 3,000 customers, feeding back algorithm optimization, forming a 'data-algorithm-scenario' positive cycle.

Tencent Cloud, SenseTime, and other peers are also very competitive, but mainly focus on the remote sensing AI large model field, integrating multimodal data resources to promote the landing of scenarios such as urban planning and ecological monitoring; in addition, academic research also contributes important breakthroughs, reportedly benefiting from policy support, including Jilin University, Northeast Normal University, and other universities participating in AI large model research and development, with Jilin Province setting up a 40 million yuan special fund to support remote sensing data processing technology research.

International companies are also accelerating their layout. For example, in the satellite remote sensing track, Satellogic deploys 300 low-cost remote sensing satellites to provide daily global coverage services and cooperates with Tencent to develop a 'satellite as a service' platform; in the multimodal large model subdivision field, well-known technology giants like IBM's EarthDial model supports cross-modal, cross-temporal remote sensing data processing, and Microsoft's GPT-4V series performs outstandingly in multispectral image analysis.

Overall, the technical competition focus of the two major markets has certain differentiation, with international companies maintaining advantages in satellite constellation deployment and hyperspectral equipment development, while domestic companies are more competitive in algorithm optimization and vertical scenario landing.

**From profitability to reinvestment: the 'growth cost' on the long-distance track**

In the long cycle, although Haiching Zhiyuan has achieved profitability, the AI track it is in is a typical 'technology-intensive + capital-intensive' industry, and continuous investment is still needed in the future.

Especially the rapid iteration speed of multispectral AI technology—international companies' investment in satellite constellations and hyperspectral equipment may form technical pressure on domestic companies; while domestic peers are also accelerating the layout of remote sensing AI large models, integrating multimodal data resources. If Haiching Zhiyuan wants to maintain its lead, it needs to continue to make efforts in three aspects: first, consolidate the barriers of the 'optical sensing and computing' technology system, second, deepen the solution capabilities of vertical scenarios, and third, disperse the risk of a single market through global expansion.

Therefore, although it turned losses into profits in 2024, considering the track it is in and the necessary long-term investment, Haiching Zhiyuan may still incur losses in the future due to market competition, R&D investment, and other factors, affecting investment returns.

The prospectus shows that as of March 31, 2025, the company's cash and cash equivalents were only 35.2 million yuan, which means that Haiching Zhiyuan, which is in the expansion period, has a strong demand for external funding.

For this reason, this fundraising in Hong Kong does not mean that Haiching Zhiyuan plans to use the raised funds to enhance R&D capabilities and increase product development investment; expand production capacity, support business growth and new product development; increase business expansion efforts and accelerate global market penetration, etc.

### 相关股票

- [Gartner (IT.US)](https://longbridge.com/zh-CN/quote/IT.US.md)