By early 2026, the AI industry is undergoing a quiet yet profound paradigm shift. The "computing power arms race," characterized two years ago by stacking GPUs and trillion-parameter models, is receding. It is being replaced by a fundamental infrastructure battle centered on "data flow." Google recently announced that starting May 2026, it will officially increase peering data transmission service fees in North America, Europe, and Asia, with rates doubling in North America. Superficially, this is a network cost adjustment, but it signals a crucial shift: the strategic value of "toll fees" for data circulation is surpassing that of computing hardware itself.

At this inflection point, XUNCE (03317) has been formally included in the constituent lists of eight indices, including the Hang Seng Composite Index, the Hang Seng Composite SmallCap/MidCap Index, multiple Hang Seng Composite Sector Indexes, and multiple Hang Seng Stock Connect Hong Kong Sector Indexes. These inclusions will take effect on March 9. This AI data infrastructure company, which listed on the main board of the Hong Kong Stock Exchange on December 30, 2025, achieved eligibility for Hang Seng Index inclusion in just 45 days and is expected to be included in the Southbound Stock Connect in March. This highlights its scarcity and growth certainty. It also signifies that capital markets are re-pricing the core value focus of the AI industry chain—shifting from "who computes fastest" to "who enables data to flow stably, accurately, and swiftly."

XUNCE's Threefold Competitive Moats Currently, domestic large language model companies have achieved globally notable technological breakthroughs in artificial general intelligence. Products like the GLM series and ABAB models demonstrate powerful language understanding and generation capabilities, providing a robust engine for intelligent transformation across industries. However, as these advanced models are deployed in high-value vertical scenarios—such as financial risk control, intelligent investment research, and industrial scheduling—a common need becomes increasingly prominent: how to efficiently, accurately, and in real-time, inject professionally curated internal data—like transaction logs, customer behavior, and compliance rules—into the model inference pipeline? This is precisely XUNCE's strategic positioning and value proposition.

XUNCE has built three core capability pillars that precisely address the three key bottlenecks for AI implementation in vertical scenarios. First, the industrial refinement capability for vertical data. In an environment of abundant generic data, the truly scarce resources are enterprises' unstructured, multi-source, heterogeneous professional data. Leveraging over 300 composable functional modules, XUNCE enables "Lego-like" assembly of cross-industry data, extending from asset management into sectors like energy, telecommunications, and retail, building a cross-scenario data processing middle platform. Second, a low-latency, highly-available circulation network. The data channels built by XUNCE achieve second-level response times and 100% accuracy, meeting the demands of highly sensitive scenarios like financial risk control and industrial control. Its services have evolved from "optional features" to "business necessities," with an actual customer retention rate exceeding 90%, forming a moat of high switching costs. Finally, paradigm empowerment from offline analysis to real-time decision-making. XUNCE drives enterprises to shift from post-event analysis in the ERP era to real-time decision-making based on data streams. In mobile advertising scenarios, it has compressed optimization cycles from monthly to daily, significantly improving ROI, which intuitively demonstrates the business value reconstruction brought by data flow.

These three pillars collectively form XUNCE's full-stack capability闭环. When the industry no longer needs just computing power or models alone, but an integrated solution that maximizes the release of data value, XUNCE's niche value becomes unmistakably clear.

Symbiosis with the Large Model Ecosystem Against this backdrop, XUNCE exhibits a deeply synergistic relationship with major large model providers. Technologically, companies like Zhipu AI and MiniMax provide the foundational cognitive engines, while XUNCE supplies the high-quality, vertical-specific data fuel. The company explicitly states it has an "upstream-downstream cooperative relationship" with model firms, forming an industrial division of labor where "models focus on algorithm optimization, and XUNCE focuses on data supply." Commercially, both parties jointly promote the deployment of AI in industry scenarios. Model providers lower the barriers to industry entry through XUNCE, while XUNCE consolidates its data channel position by supporting multi-model adaptation. This collaborative model is substantively shifting AI value realization from "technology capability demonstration" to "business outcome delivery."

This ecosystem synergy is fostering new business paradigms. While traditional annual SaaS subscriptions currently form the bulk of the company's revenue, it has begun piloting models like "revenue sharing based on performance" and future plans for "charging based on large model usage/Token consumption." Once vertical models are deployed at scale on the client side, XUNCE's revenue is expected to grow naturally with the intensity of AI usage, creating a virtuous cycle characterized by high stickiness, high margins, and high retention.

Financial performance underpins this long-term logic. From 2022 to 2024, the company achieved a revenue compound annual growth rate exceeding 124%, with gross margins consistently above 70%, demonstrating strong profitability and pricing power. As it expands into new industries and deepens its data services, growth momentum is expected to continue.

Notably, XUNCE's strategy extends beyond the domestic market. In 2026, the company plans to expand overseas alongside leading clients, establishing local teams in Singapore, Japan, and Europe. It is also venturing into cutting-edge areas like robotic training data platforms, real-time computation for commercial satellites, and aviation intelligent scheduling, continuously widening its moat.

Summary In summary, the competitive focus in the next phase of AI has clearly shifted—the most valuable asset is no longer hardware, but control over the full-stack pricing power of data flow. Recognized by the market as the "first Data Agent stock" and the "Chinese version of Palantir," XUNCE, with its solid fundamentals and highly visible growth path, has become a scarce asset in the AI infrastructure space, combining certainty with high potential. Market recognition is continuously materializing: this week, XUNCE's share price surged to HKD 78, reaching a new high since its listing, with a weekly cumulative increase of 35%. Concurrently, the broader AI large model sector was active, with leading stocks like Zhipu AI and MiniMax performing strongly. XUNCE, as a core enabler of underlying data synergy, moved in tandem with the sector, underscoring its high linkage with the large model ecosystem and further validating the capital market's strong endorsement of its positioning as the "data foundation for the vertical model era."

With its imminent inclusion in the Stock Connect, the attraction of XUNCE for "Northbound water" (mainland capital) is self-evident. Looking ahead, sustained inflows of southbound capital are expected to further broaden the company's investor base and significantly improve liquidity. This will inject new momentum into its valuation framework, accelerating value discovery.

It is noteworthy that Deutsche Bank recently issued a research report explicitly giving XUNCE a "Buy" rating and setting a target price of HKD 85. This implies nearly 20% upside potential from the current price of approximately HKD 71.4, highlighting the international investment bank's strong recognition of its long-term value.

Standing at the critical inflection point where AI transitions from "model competition" to "data collaboration," XUNCE's long-term potential undoubtedly warrants significant attention.