As China’s daily Token consumption surpasses 140 trillion, OpenAI processes 15 billion Tokens per minute. China’s Token usage grows 1,400-fold in just two years—Token, a technical term still unfamiliar two years ago, is becoming the new “kilowatt-hour” of the AI era.

 

On April 12, Shenzhen Xunce Technology Co., Ltd. (3317. HK) signed a strategic cooperation agreement with the Shenzhen Data Exchange. At the inflection point where the Token economy is moving from concept to explosion, the signing of this agreement sends a clear signal: China’s AI industry is shifting from a “model race” to a “data race”, from “general-purpose Tokens” to “vertical Token refining.”

 

Xunce Technology is becoming the core “Vertical Token Factory” in this historic process.

 

Token Economics: Why is Token the “Oil” of the AI Era?

 

To understand the significance of this partnership, one must first understand what a Token is.

A Token is the basic unit of information processed and generated by AI. You ask AI a question, consuming some Tokens; AI gives you an answer, generating some Tokens. One Token roughly corresponds to one or two Chinese characters. But the significance of Token goes far beyond being a mere “unit of measurement” ——it transforms AI into an economic resource that can be priced, traded, and even futures-traded, just as the “kilowatt-hour” gave electricity a price and the “barrel” gave oil a futures market.

Jensen Huang, CEO of NVIDIA, deconstructed the AI industry into a “five-layer cake”: energy, chips, infrastructure, models, and applications. The unified unit of measurement in all five layers is the Token. Huang’s definition: Token is the fundamental unit of modern AI, and the language and currency of AI.

Tokens are undergoing value stratification. The same Token used for casual chat is worth $0.01 per million; used for coding, it’s worth $200; used for legal document review, it’s worth $1,000—a difference of a hundred thousand times in value. Less than 5% of Token consumption creates over 80% of measurable value. The value of a Token is not determined by its production cost, but by what it is used for. This is the core logic of vertical Tokens.

Token Supply-Demand Imbalance: General Tokens in Surplus, Vertical Tokens Scarce

 

In March 2026, Liu Liehong, Director of the National Data Administration, officially named the Token Ciyuan and disclosed a set of data: China’s daily Ciyuan call volume has exceeded 140 trillion, an increase of over 1,400 times compared to 100 billion in early 2024. Nationwide, over 100,000 high-quality datasets have been established, with a total volume exceeding 890 PB—equivalent to about 310 times the total digital resources of the National Library of China.

 

At the same time, global Token demand is undergoing a structural inflection point: shifting from humans using AI to AI using AI by itself. The emergence of Agents has completely changed the rules of the game—it is not a chatbot, but an AI program capable of autonomously executing tasks.

 

If an enterprise deploys 1,000 Agents, each consuming 1 million Tokens per day, that amounts to 365 billion Tokens per year, equivalent to the total consumption of all human users in a medium-sized country. Agents don’t just consume Tokens; there are already experimental projects where Agents have their own accounts, autonomously take on tasks, earn income, and then use that income to purchase more Tokens.

 

The next surge in Token demand will no longer come from humans using more, but from machines starting to consume on their own.

 

But a deep contradiction is emerging: general Tokens are experiencing inflation, while vertical Tokens are severely scarce. Large language models can converse fluently, but once they enter vertical scenarios such as financial risk control, medical diagnosis, power dispatch, or robot control, general Tokens fall short.

 

What enterprises truly need are vertical Tokens refined through industry knowledge. The insufficient supply of high-quality vertical data has become the core bottleneck restricting the implementation of vertical large models.

 

Vertical Token Factory: The Core Positioning of Xunce Technology

Against this backdrop, Xunce Technology’s positioning as a Vertical Token Factory has emerged.

If a general-purpose large model is like a power plant, then Xunce Technology is the refinery—it does not produce basic Tokens but rather refines raw data from vertical industries such as finance, telecommunications, electric power, robotics, healthcare, and commercial aerospace into vertical Tokens that large models can directly and efficiently use.

With AI Data Agent at its core, Xunce Technology has built a full-chain technical system covering data acquisition, cleansing, standardization, real-time computation, and model fine-tuning. It can transform the complex, heterogeneous private data within enterprises into standardized vertical Tokens that large models can understand, invoke, and measure, all within milliseconds.

In 2025, the company’s revenue increased by 103% year-on-year to RMB 1.283 billion, the share of non-asset management business revenue rose to 80%, revenue per employee reached RMB 2.9 million, and ARPU jumped from RMB 2.72 million to RMB 5.59 million—behind these figures lies enterprises’ genuine willingness to pay for “vertical Tokens”.

General Tokens are crude oil; vertical Tokens are refined oil. Xunce Technology is that refinery.

Strategic Partnership with Shenzhen Data Exchange: Co-building the Standard for Vertical Tokens

According to the announcement, this strategic cooperation between Xunce Technology and Shenzhen Data Exchange focuses on three main directions, essentially co-building the production standard for vertical Tokens:

First, jointly expand data element and AI innovation businesses to promote enterprise digital and intelligent transformation. As a national-level data exchange, Shenzhen Data Exchange has leading expertise in data compliance circulation and assetization operations. The partnership will accelerate the journey for enterprises from data governance to AI applications.

Second, co-build a data assetization and data asset entry service system. As the policy for including data assets on balance sheets is deeply implemented, enterprise data is transforming from cost to asset. Xunce’s vertical Token refining capability, combined with Shenzhen Data Exchange’s compliance expertise, will provide enterprises with a standardized path to turn data into assets.

Third, establish a data specification system for Embodied Intelligence. This is the most forward-looking aspect. The demand for vertical Tokens from Embodied Intelligence (Physical AI) far exceeds that for large language models—robot training requires real physical interaction data; autonomous driving requires massive amounts of real-world driving data. The two parties will jointly develop a vertical Token specification system for scenarios such as intelligent robots, autonomous driving, and smart terminals, addressing the current industry bottleneck of insufficient supply of high-quality vertical Tokens for Physical AI training.

The Vertical Token Factory Stands at the Center of the Next Major Opportunity

Three distinct business models have emerged in the Token economy: pay-as-you-go (charge for Tokens used), monthly subscription (not charged per Token), and value-based pricing (charge based on the value created).

Xunce Technology is advancing its Token-based billing model, which embodies the logic of value-based pricing. Customer Value = Price per Call × Number of Token Calls × Number of Modules Used. The price per call for vertical Tokens is much higher than for general Tokens due to their higher business value content. In 2025, the company’s Token-based revenue accounted for 5%, with a target to increase it to 20%-30% in 2026.

Conclusion: Xunce Technology’s strategic partnership with Shenzhen Data Exchange is a micro-level manifestation of this macro-narrative. Where policy dividends, industrial demand, and technological capability converge, the Vertical Token Factory is no longer a cost center but a value engine.

As the Vertical Token Factory, Xunce Technology’s long-term value may have just begun to materialize.

Because what is truly scarce is not the Token itself, but the ability to turn every Token into refined vertical oil.

Enterprise-grade vertical Token factories and vertical models are standing at the center of the next major opportunity.