Yiyan Business Observer

This is not only user education Go with the flow (shùn shì ér wéi, going with the flow) but also a meticulously calculated land grab.

Recently, the three major operators have directly added Tokens to phone bills, causing a stir in the tech circle. This signifies that the telecommunications industry has finally reached out to the legendary 'second growth curve'—officially transitioning from selling voice and data to selling AI computing power. However, this is just the beginning: while the packages are available and prices are public (gōng kāi, public), the closed-loop of frontline services, cultivation of user habits, and integration with the model ecosystem still require time to provide answers.

On and around World Telecommunication Day on May 17, 2026, the three major operators made frequent moves: China Telecom took the lead at the group level to launch a nationwide trial commercial Token package; China Mobile introduced attractive trial prices in Shanghai and other places; China Unicom offered large free testing quotas for OPCs (One-Person Companies). The timing is quite intriguing—the public still has reservations about whether AI should be paid for separately, yet the three major operators almost simultaneously launched a combination of 'low-threshold Token packages + phone bill integrated payments,' directly reducing high-end AI computing power to a standardized product similar to data packages. This is not only user education Go with the flow (shùn shì ér wéi, going with the flow) but also a meticulously calculated land grab.

01 Data Doesn't Lie

Why now? Data doesn't lie. The window for land grabbing is always opened by an explosion in demand. The National Data Bureau disclosed astonishing figures: China's average daily Token calls were about 100 billion in early 2024, surging to 100 trillion by the end of 2025, and breaking through 140 trillion in March 2026—a more than thousandfold increase in just two years.

Against this macro backdrop, China's AI large model market size reached 49.539 billion yuan in 2025 and is expected to exceed 70 billion yuan in 2026. As Token consumption expands exponentially, the market urgently needs three things: standardized units of measurement, trustworthy settlement channels, and a unified portal for cross-model scheduling. This happens to be where operators excel—they know exactly how to package non-standard capabilities into retailable commodities and seamlessly integrate them into the billing systems of their billions of existing accounts.

02 Why Are Operators the First to 'Enclose'?

Essentially, this is an 'enclosure' based on asset endowments—not of land, but of AI capability distribution rights.

First, heavy asset preparations are complete. The three major operators collectively hold nearly 50% of the domestic IDC market share. In terms of total intelligent computing power, China Mobile reaches 92.5 EFLOPS, and China Telecom 91 EFLOPS. Over the past three years, their cumulative investment in computing power Capex has exceeded 100 billion yuan. If these massive intelligent computing centers are only used for cloud resource leasing, asset turnover rates would be too low. Tokenized retailing can transform 'idle inference computing power' into sustainable and predictable revenue, fundamentally improving the return logic of heavy assets.

Second, they are the only players holding 'accounts + authentication + clearing and settlement + national backbone networks.' The hardcore capabilities of unified API gateways, real-time metering, and streaming billing have been honed by operators for two decades. For government and enterprise clients, unified access, auditing, and billing through operators can significantly reduce procurement friction compared to dock (duì jiē, interfacing) with multiple model vendors separately.

Finally, and most importantly, the pressure from stagnating traditional businesses. In 2025, the overall revenue growth rate of the three major operators was only about 0.6%, with 'data traffic operations' seeing diminishing returns. Therefore, Token operations have been written into their core strategies. Evolving from pipeline operators to computing power service providers is a role upgrade opportunity operators have waited two decades for.

It's worth mentioning that currently, operator Token platforms mainly connect to mainstream applications like DeepSeek, Zhipu GLM, Alibaba Tongyi, Kimi, and operator-developed models. Tencent Yuanbao (Hunyuan Large Model) has not yet appeared on the Token platform access lists announced by the three major operators, possibly due to its C-end products' insistence on independent development.

03 New Challenges for Large Model Vendors

When operators become the 'toll gates' and 'showcases' for Token circulation while possessing such powerful self-developed models, it presents both immense opportunities and severe tests for third-party large model vendors:

First is the partial transfer of distribution rights: In the future, users may only remember 'I bought a Token package from China Telecom' rather than 'which model I invoked.' If the platform enables intelligent routing, individual model vendors' brand perception may be diluted.

Second is the lengthening of data feedback loops: Calls routed through operator platforms mean original call logs will predominantly settle on the operator side, requiring model vendors to redesign data reflow (huí liú, feedback) and effectiveness evaluation mechanisms.

Third is the 'acting as both referee and player' competition: Facing comprehensive efforts from Jiutian, Xingchen, and Yuanjing, third-party model vendors must accept this reality and maintain their irreplaceability by deeply binding to industry scenarios and leveraging extreme differentiation in specific fields.

Of course, the positive side is also evident: entering operator channels instantly provides access to a vast network of government, enterprise, family, and SME users, especially benefiting state-owned enterprises and government scenarios with strict compliance requirements.

04 Practical Constraints on Commercialization

Objectively, Token operations are still in the early trial commercialization stage, with at least three unresolved challenges ahead:

First, frontline services are not fully connected. Unsubscription paths, invoice issuance, and consumption detail queries vary widely, with C-end packages currently carrying obvious cultivation subsidy attributes. Second, billing transparency needs improvement. Disputes caused by 'black box billing' continue, making clarity a compulsory course (bì xiū kè, mandatory course) for the next stage. Third, real growth still needs government and enterprise validation. Personal Token packages are unlikely to contribute substantial profits short-term; the truly quantifiable second growth curve depends on whether government and enterprise MaaS procurement and industry Agent solutions can form independently disclosed revenue categories.

05 Enclosure Complete, Cultivation Has Just Begun

The three major operators have completed the initial land grab of AI circulation networks through Token packages. But enclosure doesn't equal harvest. Whether Tokens can move from financial statement footnotes to core business categories depends on two things: the depth of government and enterprise adoption and the completeness of service closed-loops. It took two decades for data traffic to become standard; Tokens may only need three to five years—and this time, operators aim to be more than just pipelines. In the long run, the party controlling its scheduling and settlement networks will hold significant circulation discourse power (huà yǔ quán, discourse power) in the AI era. The three major operators have taken a crucial step; the rest must be validated by the market.

With the full outbreak (bào fā, outbreak) of the Token economy, we stand at the dawn of a new era. Peng Yaping, Chief Marketing Officer of Huawei's Operator Business Group, stated, 'Differentiated pipeline capabilities are key for operators to realize the economic value of Tokens.' Leveraging their natural endowments of networks, computing power, security compliance, and massive users, operators are fully capable of transforming from traditional information service providers to 'Token factories' in the intelligent era.

In the future, when Tokens become as fundamental to societal operations as water and electricity, their unit prices will inevitably decline with technological progress, moving toward thin margins and universal accessibility. However, the Token economy will also undergo profound stratification—Tokens for conventional computing power will trend toward infrastructure-level thin margins and high volumes, while professional scenarios with high value and difficulty will generate enormous premium spaces.

Future competition will no longer be limited to simply 'selling Tokens' but who can build a complete closed-loop of 'scenarios + data + platforms + models.' As AI agents begin deeply embedding into production processes across industries, if operators can successfully convert their 'computing-network fusion' advantages into 'intelligent service' strengths, they will reap not just a second growth curve but the most valuable 'intellectual dividends' of the digital economy era.

END