In 2026, China’s three major telecom operators—China Telecom, China Mobile, and China Unicom—unveiled a groundbreaking strategic shift: rolling out AI token subscription plans to consumers and enterprises. This marks their third core business transformation, following the shift from voice call services to mobile data, and now to AI computing power monetization. By integrating large language model (LLM) usage into traditional phone bills, operators aim to tap the booming AI demand while reversing stagnant growth in legacy communications businesses. This article dissects the rollout status of token plans, underlying market drivers, key industry data, and persistent challenges, offering a comprehensive view of China’s AI computing power payment landscape.

What Are Token Plans & Current Industry Rollout

A token is the smallest unit of information processed by AI large models, encompassing text, code, and multimodal data. Operators’ token plans mirror traditional mobile data packages, bundling fixed token quotas for monthly or one-time purchase, with costs charged directly via telecom bills. As of May 2026, the three operators show stark differences in national rollout progress, with uneven implementation across regions.

China Telecom: First Mover with Mature Offerings

China Telecom has taken the lead in full-scale commercialization. Its offline business halls in Shanghai display clear pricing tables: a one-time 250,000-token package costs 1 yuan, while the 50 million-token tier is priced at 200 yuan (130 yuan during limited-time promotions), valid for 31 days. The official app features a prominent token plan entrance, splitting offerings into two categories: personal/home plans (9.9–299.9 yuan monthly) and developer/SME plans, all supporting third-party AI tool integration. As the most advanced in execution, China Telecom has completed end-to-end promotion and sales systems.

China Mobile: Regional Trials, Inconsistent Execution

China Mobile’s token services remain in regional pilot stages. Shanghai Mobile launched a "1 yuan for 400,000 tokens" package for individual users and small enterprises in mid-May, while Guangdong Mobile rolled out Lite (40 yuan/month) and Pro (100 yuan/month) tiers. However, Shanghai’s offline outlets and national customer service (10086) confirmed no official national launch, with frontline staff lacking systematic training. Users have raised concerns about unclear after-sales support for token plan issues.

China Unicom: Enterprise-Focused, Limited Consumer Access

China Unicom’s rollout prioritizes enterprise clients. Its Shanghai branch offers 30 million free test tokens for OPC enterprise users (valid until late June), plus 50% discounts on Coding and Token plans for first-time buyers. For consumers, no national or Shanghai-specific retail packages are available, with frontline staff and customer service (10010) reporting no public offerings. The operator’s focus remains on enterprise AI scenarios rather than mass consumer adoption.

Market Explosion & Telecoms’ Transformation Imperative

The surge in AI token demand and stagnant legacy telecom revenue are the dual catalysts driving this strategic pivot. Official industry data underscores the urgency of this shift.

Explosive Growth in AI Token Consumption

In 2025, China’s daily AI token usage skyrocketed from 1 trillion tokens at the start of the year to 100 trillion by year-end, with annual cumulative consumption reaching approximately 21,100 trillion tokens. The AI market expanded rapidly: China’s LLM market size hit 49.539 billion yuan in 2025, a 49.1% year-over-year increase, and is projected to exceed 70 billion yuan in 2026. The broader core AI industry reached 1.2 trillion yuan, accounting for 16% of the global market, with over 6,000 domestic AI enterprises.

Stagnant Legacy Telecom Revenue

Traditional voice and data services have hit a growth ceiling. In 2025, China Mobile’s annual revenue reached 1,050.2 billion yuan (0.9% YoY growth), with net profit at 137.1 billion yuan (0.9% YoY decline). Its Q1 2026 net profit fell 4.21% YoY. China Telecom’s 2025 revenue was 523.9 billion yuan (0.07% YoY growth), with net profit rising only 0.52% YoY to 33.185 billion yuan.

Surging Computing Power Revenue

By contrast, telecoms’ computing power services are booming. In 2025, industry-wide computing power revenue reached 89.8 billion yuan (11.1% YoY growth), with intelligent computing services spiking 279%. China Mobile allocated 136.6 billion yuan for computing power network construction in 2026; China Telecom earmarked 25.5 billion yuan (26% YoY growth) for computing infrastructure (35% of total 73 billion yuan capital expenditure); China Unicom invested over 17.5 billion yuan in computing power (35% of its 50 billion yuan 2026 capital expenditure), with AI revenue surging 140% YoY in 2025.

Core Challenges Facing Telecoms’ AI Token Transition

Despite ambitious plans, the shift from "data pipe providers" to "computing power service providers" faces significant hurdles, from market competition to operational gaps.

Fierce Competition from Cloud & LLM Vendors

In China’s public cloud IaaS market, Alibaba Cloud leads with a 27.9% share, followed by China Telecom’s Tianyi Cloud (12.1%) and China Mobile Cloud (10.1%). Unlike telecoms’ simplified, data-plan-style token pricing, professional LLM vendors adopt transparent, usage-based billing (separate input/output token pricing). Analysts criticize telecoms’ token fees as a "black box," with unclear consumption metrics and no differentiation between model types or usage scenarios.

Uneven Frontline Execution

Field investigations reveal widespread operational gaps. While operators promote token plans heavily in marketing, most frontline sales and customer service staff lack systematic product training, leading to inconsistent user consultations. The end-to-end service loop—from consultation, purchase, to after-sales—remains incomplete, hindering mass adoption.

Mismatched User Perception

Consumers and developers accustomed to precise LLM API pricing struggle with telecoms’ all-inclusive token packages. Unlike pay-as-you-go LLM services, telecom plans bundle fixed quotas with no rollover, wasting unused tokens. This inflexibility limits appeal for power users and enterprises with variable AI consumption.

Conclusion

China’s telecom giants’ foray into AI token plans marks a pivotal milestone in the digital economy’s evolution. Driven by explosive AI demand and stagnant legacy revenue, operators are leveraging their massive user bases, mature billing systems, and nationwide infrastructure to enter the AI computing power market. However, the transition is in its early stages, with unresolved pricing transparency, operational inefficiencies, and intense competition from established LLM and cloud vendors.

As the AI industry matures, telecoms must refine token pricing, improve service clarity, and deepen AI ecosystem partnerships to compete. For developers and enterprises integrating diverse LLMs, unified API gateways simplify multi-model access, and treerouter delivers reliable, streamlined connectivity for scalable AI workflows. Ultimately, the token economy represents not just a new revenue stream for telecoms, but a foundational shift in how AI computing power is monetized and democratized.