According to a report by China Daily on March 20…
China’s leading cloud providers are adjusting their AI service pricing in response to surging demand and escalating infrastructure costs, highlighting the growing pressures of a rapidly expanding AI ecosystem.
Alibaba Cloud recently announced price hikes of up to 34% on AI computing and storage products, including its T-Head Zhenwu 810E computing cards and CPFS intelligent storage solutions. On the same day, Baidu Cloud implemented a “structural optimization” of pricing across AI offerings, citing rising costs for core hardware and network infrastructure. Tencent Cloud has also raised prices on its large model services, reflecting a broader trend among cloud providers worldwide.
Underlying the price adjustments is a dramatic increase in token consumption—the computational unit that powers AI models. Open-source AI agents, such as OpenClaw, have accelerated this trend. Token usage linked to OpenClaw, for example, surged from 80.6 billion in early February to 358 billion by early March, more than tripling within a month. Weekly token consumption on the OpenRouter API platform reached 14.8 trillion, with agentic applications driving the bulk of the growth.
Nvidia CEO Jensen Huang noted that AI agents often require repeated calls across multiple models to complete a single task, producing “orders-of-magnitude increases” in token usage. Applications such as autonomous agents, deep research tools, and code generation systems are placing unprecedented demands on compute, bandwidth, and energy, creating tight capacity constraints for cloud operators.
Analysts say this pattern signals a structural shift in the economics of AI. Enterprises must anticipate rising costs, plan for scalable compute resources, and adapt deployment strategies accordingly. Sun Zhenya, senior research manager at IDC China, warned: “Facing exponential growth in token consumption, cost and energy use will become key constraints. Enterprises need to plan ahead in terms of computing resources and model strategies.”
The surge in AI workloads illustrates a broader industry reality: as AI systems move from experimental applications to mainstream adoption, the infrastructure that supports them—cloud compute, networking, and storage—must scale rapidly, often at significant cost. For investors and developers, monitoring token usage trends and cloud pricing dynamics has become an essential part of assessing the viability and sustainability of AI-driven business models in China and globally.