China's National Data Administration recently held a symposium on the token economy, explicitly stating it will promote the development of the token economy within its work framework.

Tokens, defined as the "smallest operational unit" for large language models to process information, have been formally designated as the measurement, settlement, and statistical unit for AI services.

The National Data Administration will focus on building high-quality industry datasets and a nationwide integrated computing power network to deepen market-oriented reforms for data elements.

This development coincides with a period of accelerating declines in computing power costs.

On the policy front, the push for a national computing network aims to lower AI computing costs at the infrastructure level.

In the market, DeepSeek recently announced a permanent price reduction for its flagship model API to one-quarter of the original price, with input costs as low as ¥0.025 per million tokens, setting a new global low for large model pricing.

The combined effect of policy and market forces is accelerating the entry of token-based AI services into an era of widespread accessibility.

The institutional advancement of the token economy, coupled with the trend of declining token costs, is seen as potentially creating new commercial opportunities for AI application-layer companies.

First is the optimization of cost structures. Lower token invocation costs may help alleviate gross margin pressures for these firms.

According to its financial report, Wondershare Technology Group Co.,Ltd. (SZSE: 300624) saw its AI server expenses increase by 78.30% year-over-year in 2025, reaching ¥80.12 million.

With reduced costs for external large model calls, cost structures could improve while maintaining model capabilities.

Second is accelerated product penetration. Theoretically, lower token costs facilitate the spread of AI features to more users and scenarios.

Early this year, Wondershare launched its Agent-driven comic drama tool, Wondershare Drama Factory, which created a full-chain production loop from script and asset creation to intelligent storyboarding and post-production editing.

In its first month after launch, the compound weekly growth rate of AI credit consumption reached 63%, and it has already empowered top comic drama producers to create hit works.

Third is enhanced economic feasibility of business models. The industry trend for the AI application layer is shifting from "one-time payments" to "usage-based billing."

As token-based pricing accelerates as an industry standard and a unified settlement system is gradually established, the revenue conversion efficiency of usage-based models may further improve.

Wondershare has taken the lead in token commercialization practices, with products like Wondershare Drama Factory and Wondershare SkyCanvas AI adopting usage-based billing models tied to token consumption.

In 2025, the company's AI-native application revenue exceeded ¥130 million, a year-over-year increase of over 90%, with paying users growing by more than 100%.

The AI industry is entering a new phase where application value is being realized. The National Data Administration's integration of the token economy into its work system signifies that a unified institutional foundation for measuring AI services is imminent, potentially lowering a key barrier for application-layer companies to achieve scaled commercial closed loops.

For companies like Wondershare, which have already completed preliminary validation in AI commercialization, token-based billing models, and vertical scenarios, this may help maintain a first-mover advantage in the new industry cycle.