- The new platform lets companies access more than 100 AI models through a single interface while paying only for the tokens they use
- The initiative reflects China’s shift from selling computing power to delivering AI as a metered utility
Zhejiang’s Wenzhou on July 15 officially launched the Southeast Token Operation Center, positioning it as a regional hub for AI computing and token services across southern Zhejiang, northern Fujian and eastern Jiangxi.
The rollout comes as China explores new ways to make access to large AI models as ubiquitous as utilities such as electricity and water.
Built on China Telecom’s Tianyi Cloud infrastructure and Xingchen TokenHub, the platform provides intelligent workload scheduling, industry-specific AI services, ecosystem integration and cross-border support.
In large language models, tokens are the basic units used to process and generate information. Every AI model invocation or AI agent interaction consumes tokens, making them an increasingly important measure of AI usage.
Billable AI services
The center is designed to turn electricity, computing power, AI models, data and applications into metered, schedulable and billable AI services, a model often described as the emerging token economy.
Alongside the launch, Wenzhou introduced its unified online portal, Ciyuan Tianxia, while ecosystem partners spanning computing infrastructure, AI models and applications signed cooperation agreements.
The platform is operated by Wenzhou Ciyuan Tianxia Technology Co., a state-controlled company. China Telecom Zhejiang is responsible for providing the initial AI computing infrastructure, while China Mobile and China Unicom are participating in ecosystem development.
Its core architecture offers six unified services: a single API gateway, model routing, token metering, payment settlement, policy-based subsidies and security auditing.
Supporting 100 AI models
The platform already supports more than 100 mainstream AI models, including DeepSeek, Qwen and Zhipu AI’s GLM, allowing enterprises to integrate once and access multiple models without additional development.
Rather than requiring customers to choose a model, the platform automatically selects the most suitable and cost-efficient option for each task, according to Chairman Yu Yangwang of Wenzhou Ciyuan Tianxia Technology.
Users can choose from standardized packages, pay-as-you-go services or customized enterprise plans, while completing computing, model deployment and application delivery through a single portal.
Government-issued token vouchers can also be applied directly to reduce costs.
Enabled by cheap, clean energy
According to Sun Xiangguang, director of Wenzhou’s Data Bureau, the city is well positioned to support low-cost AI infrastructure.
Clean energy capacity reached 9.39 GW in 2025, accounting for 45.1% of total installed capacity, while the China (Wenzhou) Data Security Harbor, a local industry park, provides compliance, trusted data circulation and auditing capabilities for cross-regional token services and overseas expansion.
Demand is also rising rapidly. Token consumption by Wenzhou’s manufacturing sector jumped 2,261% quarter on quarter in the first quarter, driven by AI adoption across manufacturing, livestream e-commerce and AIGC applications.
The city is also working with the China Academy of Information and Communications Technology to pilot interconnected AI agents, allowing autonomous AI systems to communicate directly with one another and improve operational efficiency.
Why it matters for global readers
The launch illustrates how China’s AI infrastructure is evolving beyond raw computing power toward token-based AI services, where enterprises purchase AI capabilities as a metered utility rather than managing models and hardware themselves.
For international observers, particularly in emerging AI markets across Southeast Asia and the Middle East, Wenzhou offers an early example of how green energy, data governance and industrial demand can be integrated into a regional token economy, making AI services more accessible and cost-efficient for small and medium-sized businesses.
