DeepSeek V4 Open Source Fuels Token Demand Surge; Xunce Tech's ARR Soars 300% QoQ
The open-sourcing of DeepSeek V4 is triggering renewed market attention toward the large language model (LLM) value chain. Industry consensus holds that as open-source models continue to improve in capability and inference barriers keep falling, enterprise-grade AI application deployment is entering a phase of broad-based acceleration. Token consumption across scenarios including intelligent customer service, agent collaboration, knowledge base Q&A, and business process automation is poised to enter a new cycle of explosive growth.Market observers note that in this phase of accelerated LLM adoption, the primary beneficiaries are not limited to foundational model providers alone. AI application infrastructure companies capable of handling high-frequency enterprise-level calls, accumulating vertical domain data, and generating sustained subscription revenues also command significant revaluation potential.
Xunce Technology stands as a paradigmatic example of this investment thesis. With over a decade of deep cultivation in high-barrier industries, the company has built end-to-end capabilities spanning data acquisition, cleansing, standardization, real-time computation, and model fine-tuning—refining raw, heterogeneous, non-standardized data into vertical tokens that LLMs can directly and efficiently consume. This capability stack is not only technically demanding but also critically dependent on the continuous accumulation of long-term industry know-how. Having become deeply embedded in core enterprise business workflows, it remains difficult to replicate.
From the latest performance metrics, the company's operational improvements are beginning to manifest in financial results. Xunce's 2025 results show AI revenue growing 105% year-over-year, inaugurating a second growth curve. In April 2026, Annual Recurring Revenue (ARR) related to token calls surged 300% quarter-over-quarter. Token-based paid revenue has already broken through 5% of total revenue, with a full-year target of lifting this to 20%–30%.
Beyond revenue growth, the company's penetration across customer scenarios continues to deepen. Public disclosures indicate its services now span more than ten industries, including asset management, telecommunications, power, energy, robotics training platforms, commercial aerospace, and biotech. Market analysis suggests this diversified customer structure reflects deep integration with business workflows and real-time data processing demands, conferring stronger customer stickiness and commercialization certainty.
More importantly, the impact of DeepSeek V4 is being reassessed. The core implication of open-source model iteration is not simple value compression across the industry, but rather increased model supply, reduced barriers to application development, and the embedding of AI into daily business operations for a growing number of enterprises. As model capabilities become universally accessible, enterprise focus is shifting from "whether we have a model" to "whether we have scenarios, whether we have data, whether we have stable access points."
From this perspective, Xunce's investment logic is undergoing a fundamental transition.
The market previously tended to categorize the company as a "LLM concept extension," leaving it vulnerable to sentiment volatility around foundation model competition, price wars, and open-source disruption. Yet as DeepSeek V4 drives further model proliferation, the enterprise-level inference calls, knowledge queries, customer service interactions, and agent execution demands that Xunce captures are instead seeing substantial volume expansion.
In other words, the richer the model supply, the easier it becomes for enterprises to deploy AI; the more applications are deployed, the higher the value of vertical-domain calls. It is understood that Xunce's vertical token pricing has reached $10–$100 per million tokens—more than ten times that of Anthropic—and continues to rise based on specialized use cases.
If the company continues to deliver positive signals on ARR and AI commercialization conversion rates, the market's valuation anchor for Xunce could shift from "model mapping" to "enterprise-grade AI application infrastructure."
Overall, DeepSeek V4's open-source release represents not a simplistic model substitution narrative, but a systemic opportunity characterized by accelerated enterprise AI adoption, expanding token demand, and enhanced value in vertical scenarios.
With AI revenue maintaining high growth, customer structures continuing to diversify, and application scenarios deepening, Xunce Technology is increasingly viewed as one of the most direct beneficiaries of enterprise-grade token demand growth in this wave of industry upgrading driven by open-source models.
