China’s telecom operators are turning to AI Tokens package and agent services to offset weakening mobile data income, betting that costly tools such as OpenClaw can create a new growth engine as traditional traffic businesses rapidly lose pricing power.
At the 2026 China Internet Conference, China Telecom executive Luan Xiaowei said AI had become a core force in changing production relations. He said the operator would “drive the transformation of enterprise operations from traditional traffic-based operations to Token-based operations.” China Mobile also introduced Claw Message App for remote control of OpenClaw agents by text.
Traffic Growth Brings Weaker Returns
China Telecom, China Mobile and China Unicom have begun selling AI Tokens packages that combine computing and communications infrastructure with AI models. China Telecom’s plans range from 9.9 yuan for 10 million Tokens monthly to $7.34 (49.9 yuan) for 80 million, with a $4.40 (29.9 yuan) option providing 40 million.
The strategy marks a departure from selling call time, messages and mobile data. Operators control servers and network access, but Token consumption is measured inside large models, leaving much of the billing process in the hands of model developers.
Popular systems including Alibaba’s Qwen, DeepSeek, Zhipu AI’s GLM and Moonshot AI’s Kimi have greater public recognition than operators’ own models. Telecom groups are therefore largely distributors, limiting their ability to pack AI Tokens as distinctive services customers readily understand.
Still, traditional business economics makes gambling attractive. Ministry of Industry and Information Technology data showed China’s mobile internet traffic reached 395.8 billion gigabytes in 2025, up 17.3 %t. Average monthly use per customer rose 14.1% to 20.74 gigabytes.
Yet mobile data revenue fell 3.1% to US$89.7 billion (609.7 billion yuan). During the first five months of 2026, national telecom revenue declined 1.9 percent to $108.2 billion (735.5 billion yuan) even as business volume, measured at constant prices, increased 7.9%.
Operators face a market where usage climbs while price competition, discounts and number portability squeeze income. Traffic services require more network capacity but return less revenue, pushing carriers toward AI products aimed at customers with larger budgets and greater willingness to pay.
Agents Create Expensive New Demand
OpenClaw, known online as the lobster, offers operators a route to those customers. Although its online popularity has cooled down among major internet companies, telecom groups see value in connecting the agent to their networks, billing systems and customer channels.
China Mobile’s messaging service addresses two problems: users may be away from the computer running an agent, and they may overlook alerts. Commands can be sent through the phone’s messaging application, revised repeatedly and monitored while the agent completes tasks.
The commercial appeal lies in how autonomous agents work. Unlike a basic chatbot, an agent can break down instructions, plan steps, call tools, check results, correct mistakes and repeat the process. Even a simple weather request may trigger several stages, while complicated assignments can require dozens of cycles and consume large quantities of Tokens.
That cost can deter ordinary customers, but it may be acceptable to wealthy users and businesses when automated work saves staff time or supports more valuable activity. A China Daily commentary by Ding Zhuang describes agents as digital employees being applied to credit analysis, factory monitoring, equipment maintenance and medical data review.
Ding also argued that open-source access has widened participation beyond major technology companies, allowing smaller businesses, universities and research institutes to develop applications.
“The year 2026 marks a turning point: AI is moving from a ‘hype phase’ into a period of tangible value creation,” Ding recently wrote.
For operators, that transition presents opportunity and constraint. Token packages could grow faster than mobile data, but success depends on service quality, transparent billing and applications customers consider useful. Privacy, bias, safety and accountability will also require clear rules as agents gain more autonomy.
The telecom companies’ bet extends beyond another package. They are trying to move from carrying internet traffic to supplying infrastructure, access and payment channels for AI work, using OpenClaw’s heavy Token appetite to test whether that model can produce durable revenue.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Telecom sections to stay informed and up-to-date with our daily articles.
