On March 27, 2026, Hong Kong-listed Xunce Technology (03317.HK) released its 2025 annual results, delivering a standout performance that captured market attention. The company achieved full-year revenue of RMB 1.285 billion, representing year-on-year growth of 103.28%, crossing the RMB 1 billion revenue threshold. Adjusted net loss narrowed to RMB 54.85 million, a significant improvement of 33.31% from the previous year. More importantly, the company achieved adjusted net profit of RMB 50.13 million in the second half of 2025, marking its first semi‑annual profitability and confirming a clear inflection point.

In terms of growth trajectory, first‑half revenue was RMB 198 million, while second‑half revenue surged to RMB 1.087 billion, a 449.32% sequential increase. This explosive growth closely aligns with the concentrated release of enterprise demand for real‑time data as AI models entered a phase of large‑scale deployment.

As the AI industry shifts its focus from "model training" to "inference at scale," the strategic importance of data infrastructure is being fundamentally rewritten.

Behind these results lies Xunce Technology's value realization as a core player in AI‑driven real‑time data infrastructure, signaling the arrival of a new era driven by the token economy.

Financial Inflection Point: More Than Growth – A Qualitative Shift

What makes Xunce's 2025 results particularly compelling is not just the revenue growth, but the simultaneous improvement in profitability and growth trajectory.

The shift in profitability carries even greater significance. Full‑year adjusted net loss narrowed to RMB 54.85 million, a 33.41% improvement over 2024. Looking at the second half alone, the company achieved adjusted net profit of RMB 50.14 million – a sharp reversal from the RMB 105 million loss in the first half – marking its first semi‑annual profit. This demonstrates that the company's business model has successfully navigated the transition from scale expansion to profit generation.

From a financial structure perspective, as of the end of 2025, total assets reached RMB 3.172 billion, net assets stood at RMB 2.42 billion, and cash and cash equivalents reached RMB 1.084 billion, an increase of approximately 216% year‑on‑year, providing ample "ammunition" for future strategic investments. The asset‑liability ratio was only 1.51%, reflecting minimal leverage and a very solid financial foundation.

In terms of operational efficiency, revenue per employee jumped from RMB 1.22 million in 2024 to RMB 2.87 million, an increase of 135.25%, effectively doubling workforce productivity. Customer ARPU also increased from RMB 2.72 million to RMB 5.59 million, up 105.04% year‑on‑year, demonstrating a growing ability to deepen value from each client. Based on these metrics, Guotai Haitong Securities expects the company to achieve full‑year profitability in 2026.

Token‑Based Pricing: From Selling Tools to Value Sharing

With the arrival of the token era, the role of tokens is undergoing a fundamental shift – they are no longer simply a unit of compute consumption but have become a direct carrier of business value. Xunce's core capability lies in providing vertical AI solutions that act as an "external brain" for general‑purpose large models, ensuring that every token consumed generates a measurable business return.

Xunce's solutions are deeply embedded in clients' private clouds and on‑premises systems, acting as a "data hub." Token‑based pricing means the company's revenue is directly tied to clients' AI usage – the more clients use, the more the company earns.

Building on this capability, the company is accelerating the evolution of its business model. The results announcement clearly states that Xunce is transitioning from project‑based and subscription models to token‑based pricing, evolving from a traditional tool provider into an "AI Agent enabler." According to the company, token‑based revenue accounted for 5% of total revenue in early 2026, and is expected to reach 20‑30% for the full year.

From a financial perspective, this model unlocks three layers of value: first, revenue structure optimization, shifting from one‑time project delivery to recurring service revenue with significantly enhanced predictability; second, improved pricing power, as the company's revenue becomes tied to the value created for clients; and third, expanded profit margins, as project delivery becomes less dependent on headcount, creating room for margin expansion.

NVIDIA's "Token Factory Economics" framework is validating this trend at the industry level: competition in the AI inference phase is shifting from "who has more compute" to "who can generate greater business value from each token." Xunce, with its deep expertise in processing high‑quality data across vertical industries, is perfectly positioned at this critical juncture.

Strategic Depth: From Asset Management to Robotics and Commercial Aerospace

Xunce started in the asset management industry – one of the most demanding sectors in terms of data real‑time requirements and regulatory compliance – where it holds the largest market share, covering the top 10 asset management institutions in China. It was precisely in this high‑concurrency, low‑latency, strong‑consistency environment that Xunce built its full‑stack AI Data Agent technology system, covering everything from data acquisition, cleaning, and standardization to real‑time computation and large model tuning, establishing a core technological moat in millisecond‑level real‑time data processing.

In 2025, the company accelerated the replication of this capability across other industries, including telecommunications, power, energy, urban operations, high‑end manufacturing, and healthcare. The share of revenue from non‑asset management businesses increased further from 61.3% in 2024 to 79.6%, reflecting a more diversified business mix. By year‑end, the company had developed over 300 data modules, enabling rapid adaptation to different industry needs through modular combinations and significantly reducing the marginal cost of cross‑industry expansion.

More forward‑looking is the company's expansion into two frontier areas: robotics data platforms and commercial aerospace. These scenarios, which demand extreme real‑time performance and reliability, serve both as a testing ground for technical capabilities and as an early entry into high‑growth future markets.

In addition, the company plans to steadily advance its global expansion and actively establish strategic partnerships with leading domestic and international compute providers and algorithm companies, building an integrated ecosystem that coordinates compute, algorithms, and data.

Valuation Undervaluation: Value Discovery for a Token Stock in Hong Kong

In March 2026, Xunce Technology was officially included in the Hang Seng Composite Index. Once it becomes eligible for the Stock Connect program, liquidity is expected to improve significantly. Currently, the company's PS ratio is about 40x, as the token‑based business model accelerates, the company's valuation offers significant room for upside, with the potential to align with large model companies such as Minimax and Zhipu.

Against the backdrop of the 15th Five‑Year Plan, which explicitly calls for "building a new form of intelligent economy" and deepening "AI+," demand for real‑time data infrastructure in the AI inference era is only just beginning. Xunce has demonstrated the explosive potential of growth through its revenue doubling, validated the sustainability of its model through its second‑half profitability, and opened up a new valuation narrative in the token economy through its upgrade to token‑based pricing.