On May 18, 2026, Deepexi, China’s leading AI service provider for marketing and sales decision-making, officially launched its global public offering, with plans to list on the main board of the Hong Kong Stock Exchange on May 27, vying for the title of “the first listed enterprise decision-making AI agent company.” As one of the earliest players to apply AI technology to enterprise marketing and sales digital transformation, Deepexi has built a deep foundation in the “AI + Application” field grounded in technical and data intelligence, industry expertise, and business scenarios. The underlying commercial logic and investment substance of the company warrant close attention.
IPO Fundamentals: A Structural Analysis Amid Revenue Fluctuations
According to its prospectus, Deepexi is the largest independent provider in China’s marketing and sales decision-making AI application market as assessed by Frost & Sullivan (with a market share of approximately 2.6%). Its financial statements for 2023 to 2025 exhibit the highly distinctive characteristics of a company in a “strategic transformation period.”
| Financial Indicator | 2023 | 2024 | 2025 | Trend Summary |
| Revenue | RMB 611 million (≈ USD 84.86 million) | RMB 538 million (≈ USD 74.72 million) | RMB 576 million (≈ USD 80.00 million) | Benefiting from budget adjustments by customers in the internet industry and AI product penetration |
| Gross Margin | 31.2% | 27.3% | 25.5% | Affected by rising upstream media resource procurement costs and compression of high-margin businesses |
| Adjusted Net Profit | RMB 70.81 million (≈ USD 9.83 million) | RMB 21.52 million (≈ USD 2.99 million) | RMB 24.87 million (≈ USD 3.45 million) | Excluding listing-related expenses, driven by increased strategic investment in “AI agents” |
The commercial logic of Deepexi centers on the dual-axis linkage of its two flagship platforms — AlphaDesk and AlphaData — in AI marketing.
AlphaDesk (Intelligent Advertising Placement Business)
AlphaDesk is Deepexi’s proprietary AI decision-making platform for advertising placement. In essence, it uses AI algorithms to optimize the efficiency of digital advertising placement, directly addressing enterprises’ core need to acquire public traffic. According to BCC Research, its core is built around a cloud platform system and a programmatic advertising DSP platform. Through two-way data integration, advertising exposure and audience profile data are synchronized in real time to the middle platform, helping brand clients achieve highly precise targeting and retargeting within the public traffic pool.
AlphaData (Intelligent Data Management Business)
AlphaData is Deepexi’s proprietary AI decision-making platform for enterprise customer relationship management. It focuses on the governance and commercial value extraction of first-party and multi-party data, and internally encompasses a Customer Data Platform (CDP), a Data Management Platform (DMP), an automated marketing system (EMA), and an AI engine called “Holmes.” Through proprietary algorithms, it deeply binds data feature identification with automated marketing scenarios. For example, BCC-interviewed experts noted that the “Holmes” platform not only analyzes physical information such as device IDs, but is also capable of processing user tags, behavioral characteristics, and broader third-party audience data, in order to enhance the ability to identify homogeneous or precisely targeted audiences from large-scale data.
In practical application, this capability has been deployed in refined marketing operations. When the system’s SDK tracking identifies the behavioral sequence of a user who has “added to cart but not yet placed an order,” the AI model can not only trigger subsequent automated marketing in real time (such as immediately pushing a coupon or an SMS message), but can also provide differentiated coupon issuance strategies based on the consumption characteristics of different platforms. For example, Douyin tends toward low-unit-price full-reduction coupons, while JD.com leans toward high-price consumer electronics product coupons. This approach significantly improves redemption and conversion rates while avoiding wastage of resources.
Investment Highlights and Core Competitive Advantages
Extremely high customer stickiness and a major-account moat: The company serves a large number of Fortune Global 500 enterprises. The prospectus shows that in 2025, the overall terminal customer Net Revenue Retention Rate (NRR) reached 88.7%, with the core major-account retention rate maintained above 80% for three consecutive years. At a time when customer acquisition costs are elevated, the high-stickiness base of existing major-account clients represents a powerful counter-cyclical moat.
Full-chain closed loop and incremental attribution system: Compared with providers that offer only a standalone CDP or behavioral analytics module, data from BCC Research shows that Deepexi’s core advantages are reflected in its possession of core technology strategy and product capabilities that achieve one-stop full closed-loop marketing automation deployment, using AI for automated marketing strategy analysis and decision-making, and more accurately quantifying ROI conversion outcomes through an incremental attribution system. Its built-in causal chain model helps brand clients eliminate the “performance illusion” created by organic traffic and precisely quantifies the “incremental transactions” genuinely generated by marketing activities.
Alignment with the innovative “Service-as-Software” commercial paradigm: Deepexi has not adopted a traditional fixed subscription fee model. Instead, it charges fees based on agreed key performance indicators (KPIs) for each marketing campaign. In the view of Daily Economic News, this logic of “selling results rather than tools” aligns precisely with the “Service-as-a-Software” trillion-dollar company core business model recently proposed by top global venture capital firm Sequoia Capital, and allows the company to benefit from the dividends of AI model advancement.
Future Outlook
Nevertheless, it cannot be overlooked that more than 80% of Deepexi’s revenue is tied to intelligent advertising placement. This makes the company’s business highly susceptible to being classified by clients as “marketing expenditure” rather than “system procurement,” rendering it vulnerable to fluctuations in overall advertising budgets. At the same time, media resource procurement is heavily dependent on traffic platforms such as Tencent, Alibaba, and ByteDance. As the major platforms’ internally developed commercial product capabilities strengthen, the growth space for third-party MA/DSP products is being relatively compressed.
Standing in 2026, the deployment of AI technology into vertical workflows and the real productivity dividends it generates have become the core consensus across the entire enterprise service market. Digital marketing going forward will exhibit the following trends:
Leaping toward enterprise decision-making AI agents: Digital marketing is undergoing a paradigm shift from “human operation of software” to “AI agent autonomous delivery of results.” Deepexi’s key strategic focus, the Deep Agent (intelligent agent platform), has entered commercialization. By integrating mainstream large models already available on the market, it helps clients build agent systems that demonstrate customized AI capabilities oriented toward marketing scenarios. According to Securities Times reporting, as of May 11, 2026, 37 contracts had been signed, with a total contract value of approximately RMB 23.4 million (≈ USD 3.25 million), demonstrating strong incremental potential.
Coexistence of general large models and vertical small models: Although general large models have strong cross-scenario capabilities, small models trained on highly time-sensitive data from vertical scenarios remain irreplaceable in deep marketing decision-making contexts. BCC-interviewed experts believe that there are currently no mature application cases of using AI large models to achieve one-to-one personalized marketing strategy generation and execution at the scale of massive audience populations. The future norm will be to use general large models as the “brain” for front-end interaction and agent decomposition, while entrusting the underlying analysis to the vertical marketing models that Deepexi commands.
Cross-domain complex scenarios becoming the “value anchor” for third-party service providers: Although the built-in tools of major platform operators can divert single-channel demand, large clients’ rigid demand for “public and private domain linkage, cross-channel strategy coordination, and breaking down data silos” is irreversible. This creates a unique survival space for independent third-party service providers. Just as Deepexi’s strategic focus for 2026 is to further concentrate on the combination of data-side and new automation products, through deep cultivation of vertical industry characteristics and front-loaded solution design services, it is constructing a professional boundary in the midst of the giant jungle that cannot easily be replicated.
[Disclaimer]: The above content reflects analysis of publicly available information, expert insights, and BCC research. It does not constitute investment advice. BCC is not responsible for any losses resulting from reliance on the views expressed herein. Investors should exercise caution.