Chengdu Proposes 100 Million Yuan Annual "Token Voucher" Program with 2 Million Yuan Per-Company Cap

Deep News
6 hours ago

Chengdu is set to roll out a new policy tool aimed at fueling the artificial intelligence sector, with the "Token Voucher" initiative now open for public feedback.

The Chengdu Token Voucher Implementation Plan (Trial) (Draft for Comments) was released on September 11 for a 30-day public consultation period. The plan outlines that a single entity can receive a maximum subsidy of 2 million yuan, with annual citywide token voucher issuance capped at 100 million yuan. For small and micro enterprises, the subsidy can cover up to 50% of actual purchase costs, while other entities can receive up to 30%.

Why token vouchers matter for AI development

Token, the core metric behind this initiative, is a professionally standardized term defined by the National Data Administration and the National Committee for Terminology in Science and Technology. It represents the smallest calculable unit used by large models to process multi-modal information such as text, code, and images, serving as a key value anchor in the intelligent era and the unified billing unit for large model services. Tokens are measurable, priceable, and traceable.

A token voucher is a policy tool funded by municipal fiscal resources. It supports market players in conducting lightweight AI operations such as large model invocation, intelligent agent development, multi-modal reasoning, and model fine-tuning, using token consumption as the sole accounting metric. According to a representative from the Chengdu Municipal Economic and Information Technology Bureau, the explosive growth of token usage is a clear indicator of AI's transition from the laboratory to real-world industries.

The policy rationale is threefold: first, it responds to the new trend of industry iteration by tackling the high cost of token usage; second, it strengthens the "computing power plus model" policy framework through coordinated voucher programs; and third, it seizes a first-mover advantage in institutional innovation, allowing businesses to enjoy policy benefits immediately.

Eligibility and scope of support

The vouchers will be jointly issued by the Chengdu Municipal Economic and Information Technology Bureau and the Chengdu Municipal Finance Bureau. Eligible entities include enterprises and public institutions, one-person companies (OPCs), OPC community operators, universities, and research institutions that are legally registered, operate in compliance with laws, and use tokens. Applicants must also have a good credit record over the past three years (or since establishment if less than three years), with no major safety incidents and no presence on any serious breach-of-trust lists.

The vouchers can be used for compliant token services including large model API calls for scientific research, technology development, content creation, and industrial innovation, as well as intelligent agent development and application, model fine-tuning and training services, and other AI-related innovative applications. The actual voucher amount will be determined based on the annual budget, application status, and verification results of business authenticity.

Two application models for different scenarios

The token service market in China has already developed a multi-tier structure with diverse suppliers. Comprehensive cloud providers such as Volcano Engine, Alibaba Cloud, and Baidu AI Cloud, along with professional AI model companies, telecommunications operators, third-party aggregation platforms, and vertical industry service providers, form the five main supplier categories. The comprehensive cloud giants, with their robust computing infrastructure and mature MaaS platforms, currently dominate a substantial portion of the enterprise-level token invocation market.

Service institutions will be classified into two categories, A and B, based on an expert review conducted by the Chengdu Municipal Economic and Information Technology Bureau. Entities purchasing token services from A-class institutions do not need to apply for vouchers directly; the institutions will submit recommended lists quarterly. Meanwhile, entities purchasing from B-class institutions can apply through a dedicated platform within a specified timeframe after the bureau issues application notices.

For claims, service institutions submit quarterly materials for verification. Following expert review and approval, funds are disbursed to the institutions. A-class institutions must credit the corresponding voucher value to the purchasing entities' accounts within 10 working days of receiving fiscal funds and report the disbursement details within 3 working days. The voucher value cannot be withdrawn as cash and must be used within three months, after which it becomes void and must be returned to the fiscal authorities through the original channel. For B-class institutions, purchasing entities can directly deduct their voucher value from service fees, with the institution bearing the deducted amount upfront.

Vouchers are strictly non-transferable, non-giftable, and non-tradable, except for OPC community operators who may purchase vouchers only for entities on a specified list and are prohibited from reselling at a markup. In cases where an entity cancels or refunds services during voucher use, the redeemed voucher value will not be refunded. Additionally, the same token usage cannot benefit from both provincial and municipal voucher policies simultaneously.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Most Discussed

  1. 1
     
     
     
     
  2. 2
     
     
     
     
  3. 3
     
     
     
     
  4. 4
     
     
     
     
  5. 5
     
     
     
     
  6. 6
     
     
     
     
  7. 7
     
     
     
     
  8. 8
     
     
     
     
  9. 9
     
     
     
     
  10. 10