On June 18, DataStory AI Technology Co., Ltd. (hereinafter referred to as DataStory) officially submitted its IPO application to the Main Board of the Hong Kong Stock Exchange, with CITIC Construction International serving as the sole sponsor.

New Finance News learned that just two days before the submission on June 16, the company announced the completion of a hundred-million-yuan Pre-IPO strategic financing round, with its post-investment valuation soaring to 5 billion yuan. As the third-ranked native AI enterprise in China's enterprise-level large model-driven commercial growth market, DataStory, adorned with multiple accolades such as "backed by Lei Jun's capital" and "first-tier in vertical large models," quickly became a hot target in the Hong Kong stock AI IPO race.

However, controversy has shadowed its popularity. Multiple issues have emerged, including financial data, equity operations, and compliance risks disclosed in the prospectus: Persistent losses starkly contrast with high revenue growth, gross margins have declined for three consecutive years, 866 million yuan in preferred share redemption liabilities hang like a sword of Damocles, the rationale for pre-IPO financing to inflate valuations is questionable, and strange operations such as "selling low and buying high" have appeared at the shareholder level.

As the capital market's valuation logic for AI enterprises gradually shifts from "storytelling" to "profit realization," is DataStory's hasty IPO a capitalization milestone for technological growth or a countdown to capital exit?

Three Structural Concerns Behind Increasing Revenue Without Increasing Profits

Founded in 2015, DataStory initially positioned itself as a one-stop big data and AI application provider. After continuous investments from Shunwei Capital and Xiaomi Group starting in 2021, it gradually reinforced its label as a "native AI technology company," defining itself as a "provider of AI application products and solutions for enterprise growth."

From a business architecture perspective, DataStory has built a four-layer technology stack: "bottom-layer data infrastructure—vertical large models, multi-agent systems—commercial applications." The bottom layer is a data foundation processing massive amounts of multimodal heterogeneous data daily; the middle layer consists of its self-developed SocialGPT vertical domain large model and EnlightAI multi-agent system; the upper layer implements solutions for three commercial scenarios: insight, marketing, and sales.

According to data from CIC Consulting, DataStory ranks third in China's enterprise-level large model-driven commercial growth market by 2025 revenue, with a market share of 5.8%. This industry position is repeatedly highlighted as a core selling point in the company's prospectus. However, it is worth noting that the "enterprise-level large model-driven commercial growth market" is a highly niche, self-defined sector. When placed in the broader context of enterprise-level AI marketing and data intelligence, the company's market position and competitive barriers require re-examination.

The financial data presented in the prospectus is full of contradictions: On one hand, revenue is growing rapidly; on the other, losses are expanding, gross margins are declining, and cash flow is continuously bleeding. This financial characteristic of "scale expansion but quality deterioration" constitutes the primary source of market controversy.

From 2023 to 2025, DataStory's total revenue was 235 million yuan, 283 million yuan, and 503 million yuan, respectively, with a compound annual growth rate of 46.4% over three years; revenue in the first quarter of 2026 was 79.647 million yuan, up 53.8% year-on-year, with acceleration in growth. Judging solely from the revenue curve, the company is in a high-expansion phase.

However, profitability performance strongly diverges from revenue growth. From 2023 to 2025, net losses were 115 million yuan, 86.911 million yuan, and 51.728 million yuan, respectively. While the loss amplitude (magnitude) has narrowed annually, the company has yet to cross the breakeven point. More alarmingly, the loss trend reversed sharply in the first quarter of 2026, with a net loss of 144 million yuan for the quarter, up 402.5% year-on-year, exceeding the full-year loss of 2025 in a single quarter.

The prospectus attributes the first-quarter loss surge primarily to changes in share-based payments and redemption liability interest. However, this explanation fails to fully dispel market doubts—after excluding non-operational financial factors, is the company's core business profitability truly improving? In fact, even after adjustments, the company achieved only a meager profit of 7.283 million yuan in 2025, with an extremely fragile profit base.

More concerning than losses is the continuous decline in gross margins. From 2023 to 2025, the company's comprehensive gross margins were 57.2%, 52.2%, and 42.1%, respectively, falling by more than 15 percentage points over three years; in the first quarter of 2026, they further dropped to 39.9%, nearing the gross margin levels of traditional software integrators and far from the high-margin SaaS enterprise image.

The root cause of the declining gross margins lies in the revenue structure. DataStory's revenue consists of two parts: high-margin standardized subscription products and low-margin customized solutions. From 2023 to 2025, solution revenue surged from 120 million yuan to 390 million yuan, with its share of total revenue climbing from 51.4% to 77.5%; meanwhile, standardized product revenue share shrank from 37.6% to 21.1%, and in the first quarter of 2026, it fell 8.5% year-on-year, with growth momentum nearly exhausted.

The core reason customized solutions drag down margins is that these projects require significant procurement of external services such as KOL resources, marketing placements, and third-party data, keeping outsourcing costs high. This growth model of "scaling through projects" essentially sacrifices profit quality for revenue growth, becoming increasingly unprofitable as it grows, falling into a typical "diseconomies of scale" dilemma.

A deeper issue is that customized projects rely heavily on human delivery, making it difficult to achieve the decreasing marginal cost effect of the SaaS model. As projects multiply, the company must continuously expand its delivery team, driving rigid cost increases and leaving little room for profit improvement. This explains why, despite doubling revenue growth, profits remain elusive.

Cash flow, which better reflects true operational quality than the income statement, tells a more revealing story. From 2023 to 2025, DataStory's net cash flow from operating activities was -28.788 million yuan, -41.186 million yuan, and -72.757 million yuan, respectively, negative for three consecutive years with widening outflows. The faster revenue grows, the more cash bleeds—this divergence typically indicates aggressive revenue recognition policies or poor accounts receivable recovery efficiency.

As of the end of March 2026, the company had only about 66.17 million yuan in cash and cash equivalents, while net current liabilities stood at 702 million yuan. In a state of continuous cash burn and negative operating cash flow, the company's liquidity safety margin is already quite thin. This partly explains why DataStory rushed to complete Pre-IPO financing before submission—to some extent, it resembles a lifesaving round rather than icing on the cake.

The IPO Compulsion Mechanism Amid 866 Million in Debt

Among all financial risks faced by DataStory, preferred share redemption liabilities are the most rigid and urgent, serving as the key to understanding the true motives behind this IPO.

The prospectus reveals that from 2017 to 2022, DataStory completed nine financing rounds from Pre-A to C++, raising approximately 448 million yuan in total. Due to redemption clauses commonly attached to early investment agreements, these preferred shares were accounted for as liabilities, forming a massive redemption liability. From the end of 2023 to the end of March 2026, redemption liability balances were 753 million yuan, 803 million yuan, 853 million yuan, and 866 million yuan, respectively, climbing annually and far exceeding the company's net assets.

The particularity (uniqueness) of this liability lies in its nature: It is not ordinary operational debt but capital debt with strong gamble-like terms. According to investment agreements, if the company fails to complete a qualified listing within the agreed timeframe, investors have the right to demand full redemption of preferred shares plus interest. In other words, the success of the IPO directly determines whether this 866 million yuan debt will trigger rigid repayment.

From a balance sheet perspective, as of the end of March 2026, the company had net liabilities of 681 million yuan and net current liabilities of 702 million yuan, with an extremely high asset-liability ratio. While preferred shares will automatically convert to ordinary shares upon listing, eliminating book liabilities in one fell swoop, before listing completion, this remains a sword of Damocles hanging over the company. If the IPO process stalls, fails listing approval, or raises less than expected, triggering the redemption mechanism would immediately place the company under immense funding pressure.

This "do-or-die" capital structure gives the IPO a strong "self-rescue" attribute. Some market views suggest that DataStory's primary goal in rushing for a Hong Kong stock listing is not to raise funds for business expansion but to resolve the rigid debt pressure from preferred share redemptions and provide exit channels for early investors. From this perspective, the 5 billion yuan Pre-IPO valuation serves more as an anchor for IPO pricing rather than a market-based valuation strictly based on fundamentals.

How Inflated Is the 5 Billion Yuan Market Cap Bubble?

The 5 billion yuan post-investment valuation in the Pre-IPO round is another focal point of market controversy. Is this valuation level a true reflection of value or a bubble blown by capital? Let's dissect it from multiple dimensions.

Based on 2025 revenue of 503 million yuan, DataStory's current static price-to-sales (P/S) ratio approaches 10x. While this may not seem excessive in the primary market, it appears significantly high compared to similar listed companies in Hong Kong.

Comparable Hong Kong-listed enterprises such as MiningLamp Technology-W, Fourth Paradigm, and Deepexi Technologies currently have P/S ratios generally in the 3-6x range. Among them, MiningLamp, as the industry leader, has a far larger revenue scale than DataStory in 2024, with deeper technological accumulation and a broader customer base, yet its valuation level shows no significant premium. If measured against the average valuation of Hong Kong's AI sector, DataStory's reasonable valuation range would be roughly 1.5-3 billion yuan, showing a clear gap with its current 5 billion yuan valuation.

Of course, AI sector companies often enjoy valuation premiums, but these premiums require technological barriers and growth certainty as support. With a 5.8% market share, continuously declining gross margins, and an unverified profit model, can DataStory justify a valuation hundreds of times its revenue? Market opinions are deeply divided on this.

A noteworthy part of DataStory's valuation comes from the "Lei Jun ecosystem" brand endorsement. Since 2021, Shunwei Capital and Xiaomi Group have made multiple rounds of continuous investments, totaling about 300 million yuan, becoming important institutional shareholders. At a 5 billion yuan post-investment valuation, Lei Jun's ecosystem holdings are worth about 737 million yuan, with paper profits exceeding 400 million yuan and an investment return rate of 146%.

While celebrity capital endorsements do bring brand halo (aura), industrial resources, and market confidence to enterprises, they can also easily cause valuations to detach from fundamentals. Market concerns lie in whether, after the "Lei Jun concept" hype fades post-listing, the valuation may face pressure to return to fundamentals. Especially since Xiaomi itself has extensive AI layout (deployments), the actual synergistic effect between DataStory and the Xiaomi ecosystem—and whether deep binding exists—currently lacks substantive business implementation evidence.

Another intriguing detail is the financing rhythm. After completing its C2 round in 2022, DataStory experienced a four-year financing hiatus, only to urgently complete a Pre-IPO round two weeks before submission, raising 78.86 million yuan from Jingjiang Yatai, Xinjilinghang, and Yuanguzhou for new share subscriptions.

From an effect standpoint, this financing round held more symbolic than capital significance—less than 80 million yuan in funds is a drop in the bucket for a company with 500 million yuan in annual revenue, but its primary role was to anchor IPO pricing at a 5 billion yuan reference. This "last-minute" valuation boost raises questions: Is it market-based pricing or a deliberate arrangement to inflate the IPO offering price? Market interpretations vary.

Information Disclosure Flaws and Gray Areas in Shareholder Operations

Beyond financial and valuation controversies, DataStory has also exposed multiple suspicions in corporate governance and information disclosure, reflecting issues with sponsor work quality and internal control rigor.

Most attention-grabbing is the abnormal operation by shareholder Yuanguzhou. The prospectus reveals that during the Pre-IPO round capital increase, Yuanguzhou and Shanghai Aizhirui cashed out 9.1492 million yuan and 13.5508 million yuan, respectively, through old share transfers, with transfer prices discounted by 63.5% compared to the same period (concurrent) capital increase price.

Public records show that Yuanguzhou's executive partner, Sun Xiaotian, simultaneously serves as DataStory's co-secretary, board secretary, and investment and financing director. In other words, the investment institution controlled by the company's board secretary engaged in low-price old share sales while participating in new share subscriptions at high prices. This abnormal operation of the same entity "buying high and selling low" at the same timepoint defies normal business logic.

Market questions arise: Why such a large price discrepancy between old share transfers and new share capital increases in the same financing round? Second, as an insider, does the board secretary's use of affiliated institutions for simultaneous buying and selling involve interest transfer or shareholding proxy arrangements? Third, does this operation involve interest transfer through old share transfers or conceal certain undisclosed equity arrangements at low prices?

More surprisingly, the prospectus contains a basic error in disclosing a shareholder's full name. Yuanguzhou's full name is "Fuzhou Yuanguzhou Venture Capital Partnership (Limited Partnership)," but the prospectus incorrectly listed it as "Fuzhou Yuanguzhou Equity Investment Fund Partnership (Limited Partnership)," changing "venture capital" to "equity investment fund."

Following standard IPO due diligence procedures, all shareholders' business licenses and partnership agreements should be archived and verified as base documents, with full shareholder names being the most fundamental disclosure information. Such a basic error suggests obvious loopholes in the sponsor CITIC Construction International's due diligence and document review processes. This raises concerns: If even shareholder names can be miswritten, might more complex financial data, business descriptions, and risk disclosures in the prospectus contain similar oversights?

Notably, shortly before DataStory's submission, at the Hong Kong Stock Exchange's Future Technology Summit, Liu Ying, co-head of the IPO review department, publicly criticized some prospectuses for "flowery language, unclear business model descriptions, beautified industry rankings, and vague revenue recognition methods," even Speak frankly (bluntly stating) that "some prospectuses read like advertisements from page one to five." Regulatory statements form a subtle contrast with the issues exposed by DataStory.

The "industry third" ranking also faces suspicions of "customized sector" beautification. Citing CIC Consulting data, DataStory claims to rank third in China's "enterprise-level large model-driven commercial growth market" with a 5.8% market share. However, this sector definition is extremely narrow—limiting to "large model-driven" and "commercial growth" effectively carves out a tiny niche from the broad enterprise-level AI market.

If placed in larger sectors such as "enterprise-level marketing technology" or "data intelligence software," competitors like MiningLamp and percentage point ( percentage point 科技) significantly surpass DataStory in revenue scale and customer coverage, pushing the company's industry ranking much lower. This tactic of carving out a highly specific sector to achieve a top ranking is not uncommon among AI companies going public, but it often draws market criticism as "ranking high in a tiny pond."

Double Red Lines of Data Collection and AI Supervision

As an AI company centered on public social data as its core means of production, data compliance is an unavoidable risk topic for DataStory and a deep-seated factor of uncertainty for its IPO.

The foundation of DataStory's business lies in the massive collection of public UGC data from social media and e-commerce platforms, which is then processed through algorithms to generate business insights. This business model naturally operates in a sensitive area of data regulation. The Cybersecurity Law, Data Security Law, Personal Information Protection Law, and the Regulations on the Administration of Network Data Security, which will come into effect in 2025, have established an increasingly stringent data regulatory framework.

The risks are evident in the boundaries of collecting public data. Although the company emphasizes that it collects only public information, social media data often contains personally identifiable information. Whether large-scale crawling and commercial use are fully compliant remains a gray area. In recent years, several big data companies have been penalized for non-compliant data crawling, and regulatory red lines continue to tighten.

The Interim Measures for the Administration of Generative AI Services introduced in 2023 require providers of generative AI services to fulfill obligations such as algorithm filing, security assessments, and content reviews. DataStory's SocialGPT large model is used in business analysis scenarios. Although it does not directly target C-end users, it must still comply with relevant regulatory requirements, leading to continuously increasing compliance costs.

Going public in Hong Kong may itself trigger data security reviews. According to the Measures for Cybersecurity Review, network platform operators handling the personal information of over 1 million users must undergo cybersecurity reviews when seeking overseas listings. Although DataStory primarily serves B-end enterprises, its underlying data originates from the public content of a vast number of C-end users, creating uncertainty about whether it will reach the review threshold.

While the prospectus lists data compliance as a risk factor, it does not provide detailed disclosure of the company's proof of data source compliance, specific compliance measures already taken, or whether it has faced relevant penalties in the past. Against the backdrop of increasingly stringent data regulation, this risk has a "black swan" attribute—once triggered, it could have a disruptive impact on the business.

The Debate on the Survival Space of Vertical AI Vendors

From an industry perspective, the most significant challenge DataStory faces is not financial or valuation-related but rather the survival value of vertical large model vendors in the era of general-purpose large models.

Currently, the competitive landscape in the enterprise-level AI marketing sector is being reshaped. On one hand, traditional marketing technology leaders such as Mininglamp Technology and AdMaster have deep customer accumulations, more complete product matrices, and stronger delivery capabilities. On the other hand, tech giants like Baidu, Alibaba, and Tencent are leveraging general-purpose large models to penetrate downward, rapidly entering commercial application scenarios with their advantages in computing power, algorithms, and ecosystems. The rapid advancement of general-purpose large models is continuously compressing the differentiation space for vertical large models.

DataStory's core competitive barrier lies in its data accumulation and scenario understanding in vertical industries such as fast-moving consumer goods. However, this barrier is not insurmountable. When general-purpose large models can achieve effects close to those of vertical models through fine-tuning, why would customers still pay a premium for vertical solutions? This is the existential question that all vertical AI vendors must answer.

A more pressing issue is the intense homogenized competition in the industry, with price wars escalating. To secure top-tier clients, service providers are lowering project bids and even accepting losses to secure orders. DataStory's declining gross profit margin largely reflects this industry-wide price war. Without exclusive data sources, monopolistic algorithms, or irreplaceable products, a 5.8% market share is insufficient to form a robust competitive moat.

Additionally, enterprise clients, especially Fortune 500 companies, possess strong bargaining power and often demand price reductions or additional services during contract renewals. DataStory's top five clients account for approximately 26% of its revenue. While customer concentration is not extremely high, the loss or price pressure from key clients could significantly impact performance. In a buyer's market, the company's bargaining power and customer retention capabilities face challenges.

DataStory's IPO is less a natural progression of capital accumulation and more a critical battle amidst multiple pressures. The time window for preferred stock redemption, ongoing cash burn, and exit demands from early investors collectively push the company to go public on the Hong Kong Stock Exchange at a valuation of 5 billion.

Capital markets welcome AI companies but reject formulaic operations of "storytelling, hype, and valuation inflation." As the AI industry transitions from the conceptual phase to the implementation phase, the market's pricing logic is undergoing profound changes. Only companies that can deliver tangible profits will receive valuation premiums, while those that merely tell growth stories will gradually be marginalized. Regarding DataStory's IPO journey, New Finance News Network will continue to follow up.