Produced by | Bullet Finance

Art Editor | Qianqian

Reviewed by | Songwen

Recently, Visual China officially submitted its listing application to the Hong Kong Stock Exchange, aiming to list in Hong Kong and create a dual 'A+H' listing platform.

As a leading player in China's visual copyright sector, Visual China once enjoyed significant industry dividends through its copyright distribution model, securing its top position.

However, times have changed. Generative AI has lowered the barrier to content production overnight, reshaping content creation logic. The copyright library that Visual China once took pride in is rapidly losing value due to technology—its core business has been declining year after year, high-value clients continue to leave, and long-standing copyright disputes have never truly dissipated.

Amid internal and external pressures, the company is struggling to find a second growth engine, with high hopes pinned on its AI business. According to an incomplete count, the term 'AI' appears over 300 times in Visual China's prospectus.

Yet, the company's R&D investment has been shrinking in recent years. In 2025, its value-added services segment, which includes AI training data services, generated just RMB 45.542 million in revenue, accounting for only 5.9% of total revenue.

In the capital market, Visual China is seeking to raise funds through a Hong Kong IPO while its controlling shareholder precisely cashed out before the listing.

With its old business struggling and its new story yet to materialize, will Visual China's listing—fraught with uncertainties—mark a turning point or just another round of capital storytelling?

Founded in 2000, Visual China operates a visual content copyright trading platform under its namesake brand. By integrating images, videos, audio, and other content, it serves news media, brand companies, and internet enterprises.

The business logic primarily involves signing contracts with a vast number of photographers, designers, and professional content agencies upstream to aggregate scattered visual content into a massive exclusive or agency copyright library. Downstream, it connects with media, enterprises, advertising agencies, and other demand-side entities, charging commissions through 'licensed use without selling off copyrights.'

In essence, a sufficient and clearly copyrighted content inventory is the foundation of Visual China's business.

The decade of rapid mobile internet development was the golden age for this copyright distribution model. At that time, high-quality visual content was scarce, individual access channels were fragmented, and industry copyright rules were imperfect (imperfect). As a centralized trading platform, Visual China was both the most efficient way to find images and the safest compliant choice, naturally growing rapidly on industry dividends.

(Figure/Prospectus)

However, with the rise of generative AI, the industry's rules have been rewritten. The pace of AI technological iteration has far exceeded expectations, with new content production methods and business scenarios emerging one after another, shaking the foundation of traditional stock image businesses.

The impact was first felt on the content supply side.

In the past, a high-quality landscape or commercial image required significant investment from photographers in travel, equipment, and lengthy shooting and selection processes. Today, with a precise text description, AI can generate comparable results at a fraction of the cost and time, making the gap in cost and efficiency starkly apparent.

For cost-sensitive corporate clients, especially with generally tightened marketing budgets, such a cost-effective alternative is highly attractive.

As AI lowers the barrier to content production, the commercial value of traditional copyright licensing naturally diminishes, a change clearly reflected in Visual China's performance.

From 2023 to 2025, the company's revenue was RMB 781 million, RMB 811 million, and RMB 778 million, respectively, with growth slowing from 11.94% in 2023 to a decline of 4.03% in 2025. Net profit attributable to shareholders shrank more significantly, falling from RMB 146 million to RMB 83.0275 million over three years, with negative growth for two consecutive years and a more than 30% drop in 2025, indicating a sustained contraction in profitability.

(Figure/Visual China's 2025 Financial Report)

As a revenue pillar, content licensing services felt the downward pressure first. In 2025, revenue from this core business was RMB 524 million, down 14.1% year-on-year, with its revenue share declining from 73.7% in 2023 to 67.2% in 2025.

(Figure/Prospectus)

Visual China admitted in its prospectus that the decline was partly due to some clients tightening marketing budgets and reducing content procurement spending. Additionally, the rapid development of AI technology has indeed changed content procurement and consumption habits across the industry.

Behind this, Visual China is facing the loss of high-value clients. From 2023 to 2025, the number of key account (KA) clients spending over RMB 100,000 annually decreased from 776 to 631, with client retention rates dropping from 83.0% to 77.6%.

The decline in high-value paying clients and retention rates directly reflects the diminishing appeal of traditional copyrighted content.

(Figure/Prospectus)

More challenging than the performance decline is the persistent copyright compliance burden the company has yet to shed. In fact, Visual China's 'image-selling' business has repeatedly faced public opinion and compliance crises.

The most well-known incident was the 2019 'black hole photo copyright' controversy. At the time, Visual China faced a public backlash for claiming copyright over the 'black hole' photo released by the European Southern Observatory.

Shortly after, the official Weibo account of the Communist Youth League Central Committee posted screenshots of national flags and emblems on Visual China's website, asking, 'Does your company also own the copyright to the national flag and emblem?'

For the first time, its copyright operation model faced nationwide scrutiny. Subsequently, the company was summoned by regulators, its website suspended for service, and underwent comprehensive rectification.

Even more criticized was Visual China's 'litigation-driven sales' business model, which used web crawlers to track the usage of its images online, provided copyright protection services such as online infringement evidence preservation, and acquired clients through litigation.

In 2023, renowned astrophotographer Dai Jianfeng publicly accused Visual China of demanding compensation, sparking industry outrage. The case ended in 2025 with Visual China losing the lawsuit, compensating the opposing party for economic losses, and issuing a public apology. In 2024, a research report by Tsinghua University Law School directly criticized its 'copyright fishing' and 'litigation-driven sales' logic, categorizing this model as a 'copyright pest' controversy.

Even in its latest prospectus, Visual China disclosed an ongoing copyright lawsuit with TIYU (Beijing) Culture Media Co., Ltd., explicitly listing 'copyright compliance risks' as a major risk factor for the company.

For an IPO company, the economic compensation from a single lawsuit is just the tip of the iceberg. The longer-term impacts are damaged brand reputation and eroded client trust. As it rushes toward the Hong Kong Stock Exchange, these lingering copyright disputes will remain a core focus for the market to assess whether its business model is reasonable and sustainable.

Faced with sustained pressure on its traditional copyright business, telling a compelling new story has become crucial for Visual China to break free. This time, AI has become its focal point.

In the AI sector, Visual China's strongest asset is its accumulated content library.

The prospectus shows that the company currently has over 700 million multimodal content assets, including images, videos, audio, and 3D models, with over 70 billion multidimensional data tags, all with complete and traceable copyright licensing chains.

Amid stricter data compliance requirements, this clear copyright chain is an advantage that ordinary data annotation companies cannot easily replicate, forming the core competitiveness of its AI training data business.

Leveraging this asset, the company has begun providing training data services to leading large language model (LLM) developers, exploring new commercialization paths. It has entered the supply chains of major LLM developers such as Microsoft, Alibaba, Tencent, and MiniMax.

Currently, this business falls under the 'value-added services' segment, which generated RMB 45.542 million in revenue in 2025, up 20.4% year-on-year, but remains limited in scale.

In addition to exporting training data, Visual China has also developed its own image understanding LLM for vertical scenarios, but it is primarily used internally—such as improving content warehousing (warehousing) efficiency through AI auto-pre-annotation, enhancing text-to-image and image-to-image search experiences with CLIP architecture, and assisting copyright protection with AI duplicate detection systems.

In other words, these AI capabilities primarily optimize the efficiency of its traditional content library rather than forming independent, high-growth business lines.

(Figure/Prospectus)

Overall, whether it is training data services for LLM developers or internal efficiency-enhancing tools, Visual China's AI Layout (layout) is still in its early exploration stage, far from ready to replace traditional businesses as the core growth engine.

A more critical issue is that the scarcity of training data is not a long-term barrier. In the early stages of AI development, high-quality, clearly copyrighted datasets were indeed scarce. However, as generative technologies continue to iterate, the scarcity of content itself will diminish.

At the same time, leading LLM developers are building their own data systems. In the long run, the barriers for third-party data service providers are likely to be continuously diluted.

At this critical juncture for strengthening competitiveness, Visual China is telling stories about its AI Layout (layout) while shrinking its R&D investment.

From 2023 to 2025, the company's R&D expenses were RMB 99.061 million, RMB 72.698 million, and RMB 73.024 million, respectively. In 2024, R&D expenses fell 26.6% year-on-year and remained largely flat in 2025, showing no significant increase.

With annual R&D investment below RMB 100 million, the company is aggressively pushing its AI business Layout (layout), raising questions in the market about the necessity of its fundraising.

Even more intriguing is the company's financing pace.

Around the same time as submitting its Hong Kong IPO application, Visual China announced in March 2026 its plan to apply for the registration and issuance of up to RMB 400 million in science and technology innovation bonds with a term of no more than 10 years. The funds would be used for projects in science and technology innovation, R&D investment, mergers and acquisitions, and restructuring.

While pursuing equity financing in Hong Kong, the company is also applying for RMB 400 million in science and technology innovation bonds. For a light-asset company with annual R&D investment below RMB 100 million, such large-scale financing requires more thorough explanation of its actual use and necessity.

If the science and technology innovation bonds can cover the funding needs for AI R&D and daily operations, why would the company choose more costly equity financing? Is this driven by genuine business needs or a tendency to 'maximize financing'?

In response to these questions, Bullet Finance sent an interview request to the company, which stated it is currently in the quiet period before listing and cannot accept interviews.

More notably, Visual China's controlling shareholders significantly reduced their stakes ahead of the Hong Kong IPO.

On May 26, 2026, Visual China disclosed that its controlling shareholders, Wu Yurui and Chai Jijun, had completed the reduction of a combined 12.9498 million shares. After the reduction, the combined stake of the controlling shareholders and their concerted parties dropped from 21.11% to 19.27%. The shares reduced were all acquired during the company's restructuring and listing in 2014.

(Figure/Visual China Announcement)

Based on the stock price during the reduction period, these shareholders cashed out approximately RMB 272 million, exceeding the company's 2025 net operating cash flow (RMB 97.959 million).

Twenty days later (June 15), the company submitted its prospectus to the Hong Kong Stock Exchange, planning to raise funds through a Hong Kong IPO.

If the company truly faces a funding gap for development and urgently needs to raise funds through listing to seize AI industry opportunities, why would the controlling shareholders choose to cash out on a large scale before the Hong Kong IPO instead of sharing the development risks with the company? Their actions with real money conflict sharply with the listing narrative of 'urgent funding needs.'

Bullet Finance noted that Liao Daoxun and Wu Yurui are spouses, and their son, Liao Jie, is the company's board chairman.

Over the long term, Visual China's controlling shareholders have already achieved significant returns through multiple rounds of stake reductions. According to data from Hithink RoyalFlush, including this latest round, the three controlling shareholders have cashed out nearly RMB 1.5 billion in total.

In contrast, secondary market investors have long struggled to obtain reasonable returns. Since its A-share listing, Visual China has distributed a total of approximately RMB 208 million in cash dividends. As of now, its dividend yield is just 0.08%.

(Figure/Hithink RoyalFlush)

Now, as the company attempts to re-enter the capital market with an unproven AI narrative, it faces traditional core business stagnation, unresolved copyright compliance burdens, and controversial controlling shareholder actions before the IPO, adding more uncertainty to its listing journey.

*The featured image in the article is from Shetuwang, based on the VRF protocol.