As the AI industry enters a more mature phase, a significant shift is occurring in the capital markets' valuation logic. The focus is moving away from a singular pursuit of model capabilities and computing power scale towards a reassessment of the "Token is King" paradigm. During this transition in market perception, Xunce Technology (03317) has emerged as one of the most representative Token-related stocks in the Hong Kong market.
As a leading Chinese provider of AI real-time data infrastructure and analytics services, Xunce released its full-year 2025 results. This report not only demonstrates a leap in the company's revenue scale but, more importantly, signals the formal establishment of its "profitability inflection point" through a strong earnings performance in the second half of the year. Notably, since March 2026, the company's stock price has experienced a substantial surge in a short period. On its first trading day following inclusion in the Southbound Stock Connect scheme, the share price rose by over 70%, capturing significant market attention.
Fundamentally, this rally is anchored by "better-than-expected" 2025 results, amplified by a confluence of factors including a shift in investor base, an upgrade in its industry narrative, and a reshaping of institutional perception. In other words, Xunce is undergoing a classic "Davis Double Play": on one hand, earnings exceeding expectations are driving upward revisions to profit forecasts, while on the other, its valuation framework is being elevated due to a sector-wide reassessment.
**The Ten-Billion Revenue Mark and the H2 Surge: Core Signals of an Inflection Point** In 2025, the company achieved total revenue of 1.285 billion yuan, a year-on-year increase of 103.28%, marking its first time crossing the ten-billion-yuan threshold. This milestone is highly significant for an AI infrastructure company, indicating that its products and solutions have successfully transitioned from early-stage validation to scaled replication. However, what truly shifted market sentiment was not the "doubling" itself, but the structural change in the growth trajectory: revenue for the first half of 2025 was 198 million yuan, while second-half revenue surged to 1.087 billion yuan, representing a quarter-on-quarter increase of 449.31%.
This "back-loaded" revenue curve often carries strong signaling significance for technology companies. Correlating with the AI industry cycle, this inflection point coincides with the transition of large language models from experimental phases to enterprise-level deployment. Businesses are beginning to genuinely embed AI into their core operational workflows, driving a sharp increase in demand for data infrastructure. Deconstructing this growth reveals its essence: explosive AI demand multiplied by the scarce supply of data infrastructure, resulting in non-linear revenue expansion.
Examining the specific business segments reveals clear underlying drivers: on one hand, the technological capabilities accumulated in the asset management industry continue to provide stable revenue; on the other, diversification into multiple industries has become a new source of growth. In 2025, revenue from non-asset management businesses accounted for approximately 79.6%, an increase of 18.3 percentage points from 61.3% in 2024. This indicates the gradual realization of the company's ability to replicate its success across industries, validating the broad applicability of its AI data infrastructure.
**From Losses to Profit: Business Model Validated at Scale** More transformative than revenue growth is the improvement in Xunce's profitability. For the full year 2025, the company reported a net loss of approximately 130 million yuan, which widened compared to the previous year. However, this figure was impacted by non-recurring items amounting to roughly 74.8 million yuan. Excluding these one-off effects, the adjusted net loss was 54.85 million yuan, narrowing by about 33.41% year-on-year.
Analyzing profitability on a half-year basis reveals a clearer picture: the adjusted net loss for the first half of 2025 was 105 million yuan, while the second half saw an adjusted net profit of 50 million yuan. This successfully marks an earnings inflection point, representing the first time the company achieved positive profitability on a half-year basis. This change sends a crucial signal: Xunce's business model has crossed the breakeven point and has begun to demonstrate profitability at scale.
From an operational perspective, this inflection point stems from three main factors: First, the expansion of revenue scale has led to a dilution of fixed costs. As the company entered the ten-billion-yuan revenue bracket, the marginal pressure from R&D, sales, and administrative expenses decreased significantly. The ratio of R&D expenditure to revenue dropped markedly from 71.3% the previous year to 48.0%, indicating improved operating leverage and that R&D investments are entering a harvest period. Second, a platform-based product architecture has significantly enhanced delivery efficiency. Through modular design, the company's solutions can be rapidly replicated, reducing reliance on extensive manpower. In 2025, revenue per employee reached 2.87 million yuan, a substantial increase of 135.25% year-on-year. Third, customer value continues to rise. The average revenue per user increased from 2.72 million yuan in 2024 to 5.59 million yuan in 2025, a growth of 105.51%, reflecting its transition from a "tool supplier" to a "core system provider."
According to the patterns of technology company development, when a firm simultaneously exhibits "high revenue growth + rapidly narrowing losses + increasing customer value," it often signifies that its business model has entered a virtuous cycle. However, the company appears not to be resting on its laurels. According to the earnings announcement, Xunce will continue to deepen the evolution of its business model, accelerating the transition from project-based and subscription models towards a Token-based payment system and revenue-sharing arrangements based on performance. This shift will deeply align the company's income with the value it creates for customers, enhancing revenue predictability and unlocking significant potential for margin expansion.
Under the Token payment model, the customer price can be broken down as: price per call × number of Token calls × number of modules applied. Xunce's management stated during the earnings call that the company has adopted the Token payment model in 2026, with Token-based fees currently accounting for 5% of revenue. They expect this proportion to rise rapidly to 20-30% within 2026. This evolution towards "value-based pricing" ties the company's earnings directly to customer value creation, releasing significant potential for margin expansion. This implies that past revenue growth was linear, whereas future growth is poised to be exponential; past growth relied on stacking customer numbers and average revenue per user, while future growth will be driven by an explosion in model calls and the tangible business outcomes powered by high-quality data.
**From Data Tool to AI Infrastructure: The Core Logic Behind the Performance** Interpreting Xunce solely through its financial data risks underestimating its long-term value. In reality, the performance surge is underpinned by a fundamental change in the company's role within the AI industry chain. Xunce has built a comprehensive technical system centered around its AI Data Agent, covering data collection, cleansing, standardization, real-time computation, and large language model fine-tuning, forming a unified data infrastructure platform.
The essence of this platform is the transformation of complex internal and external enterprise data into readily usable "means of production." In the AI era, the importance of this capability is rapidly escalating. For instance, within the next-generation AI operating system OpenClaw, Xunce plays the critical role of core "data Token supplier" and "data fuel." Leveraging its ability to process multi-source, heterogeneous data in milliseconds, the operating system accesses not "yesterday's reports" but "the truth of the present moment" when calling upon data.
This shift directly leads to two outcomes: first, a significant enhancement in customer stickiness, as the system becomes the enterprise's "data hub"; second, an increase in willingness to pay, thereby driving rapid growth in average revenue per user. From an industry perspective, this signifies that Xunce is moving into a more strategically valuable position—it is no longer just a part of IT expenditure but is becoming a core component of enterprise AI capability infrastructure.
Therefore, after laying the foundation with its 2025 results, Xunce entered a window for valuation reassessment in 2026, a process driven by multiple factors. First is the direct catalyst from the investor base. On March 9, 2026, the company was formally included in the Hang Seng Composite Index and the Southbound Stock Connect scheme, allowing mainland capital to participate directly in trading. This change significantly improved the company's liquidity and prompted a valuation re-rating. The sharp share price increase on the inclusion date reflected a repricing driven by the shift in the investor base structure. Second is the fundamental support provided by the earnings beat. The revenue doubling, the second-half surge, and the emergence of the profitability inflection point prompted the market to reassess the company's growth potential and profit capacity. Third is the upgrade in the industry narrative. With the rising popularity of the OpenClaw open-source AI agent framework, Xunce launched its FeedClaw platform, targeting the data governance and Token optimization segment for AI Agents. Consequently, the market has redefined Xunce as a "leader in AI infrastructure (Token)" and assigned it the label of the "Token Pioneer Stock." Simultaneously, market narratives surrounding the "Lobster Trio" (MiniMax, Zhipu AI, and Xunce) have further strengthened its recognition within the AI sector, fueling a resonance of capital inflows and positive sentiment.
In light of this, CITIC Securities has drawn a parallel between Xunce and "the Chinese version of Palantir," emphasizing its scarcity as a real-time data infrastructure company and its potential to continuously benefit from the demand for data infrastructure driven by the AI implementation wave.
In summary, Xunce's stock price surge is essentially the combined result of "earnings validation + capital restructuring + narrative upgrade."
**Summary** Returning to the core results, Xunce's 2025 financial report has clearly answered a key question—is its business model viable? The answer is affirmative: revenue has achieved scale, the path to profitability is becoming clear, and operational efficiency continues to improve. More importantly, the performance reflects underlying industry trends. As AI enters the phase of scaled application, high-quality data is becoming a core factor of production, and data infrastructure is the key platform enabling this factor. In this context, Xunce is not merely a software company but an important infrastructure provider for the AI era. The evolution towards a Token-based payment model positions the company to potentially become a scarce strategic asset within the Token economy wave.
In the short term, the company's stock price has completed its first round of valuation repair. However, from a medium- to long-term perspective, its true potential will depend on the pace of AI application penetration and the development of the data factor market. If the AI industry's development path continues to deepen at the enterprise level, companies like Xunce that possess data infrastructure capabilities may not just be "beneficiaries" but could become among the rule-makers themselves.
Comments