Over the years, with the mobile internet evolving at breakneck speed, the transformation of telecom operators into mere data conduits has become an unavoidable trend. Against this backdrop, the three major operators have recently jointly launched token packages. Can token sales truly become a new profit avenue for operators?

I. Three Major Operators Launch Token Packages

According to Xinhua News Agency, with the rapid advancement of AI technology and the flourishing development of agent and large-scale model applications, China's token usage has soared. Recently, China Mobile, China Unicom, and China Telecom have all rolled out token packages for individual and government-enterprise users. The three major operators are racing to enter the token market, not only democratizing AI accessibility for ordinary consumers but also fostering the deep integration of communication, computing power, and AI industries, thereby creating new opportunities and spaces for digital economy development.

Data from the National Data Bureau reveals that at the beginning of 2024, China's average daily token usage stood at 100 billion; by the end of 2025, it had skyrocketed to 100 trillion; and in March of this year, it surpassed 140 trillion, marking a thousandfold increase in just two years. The demand for AI computing power continues to surge, and the three major operators are leveraging their strengths in computing power networks, cloud-network integration infrastructure, and vast user bases to accelerate the marketization of token products.

On May 17, China Telecom launched a trial commercial token package, offering multiple options for individuals and developers, with the lowest package priced at 9.9 yuan per month for 10 million tokens. On May 16, China Unicom's Shanghai branch announced token services for local OPC (One-Person Company) users, with each user eligible to receive 30 million tokens for free testing. In April, China Mobile rolled out token packages in Beijing, Hubei, Henan, and other regions...

"After a few words and tasks assigned to the agent, tens or even hundreds of thousands of tokens are quickly consumed," said Mr. Su, a Beijing resident. Many people share Mr. Su's experience, often encountering issues such as high computing costs, cumbersome model access, and fragmented payment channels when using large models, AI creation, and agent services.

II. Can Selling Tokens Become a New Profit Opportunity for Operators?

Recently, the three telecom giants—China Mobile, China Unicom, and China Telecom—have all positioned "Token Operations" at the core of their strategies, launching token tariff packages for individuals and enterprises. How should we interpret this synchronized move? How should we analyze this situation?

Firstly, shifting to selling tokens is an inevitable choice for the three major operators after their traditional businesses reached their peak. For a long time, the core profit logic of the three major operators has centered around voice, SMS, and data traffic. Especially during the 2G era, operators were the cornerstone of the mobile internet, with packages like 5 yuan for 30M serving as a key revenue stream. However, with the widespread adoption of 4G, 5G, and the impending 6G era, this model has hit its ceiling. On one hand, the unit price of data traffic continues to decline, and under the "speed up and reduce fees" policy, data traffic has transformed from a scarce resource to a common commodity, with per capita data usage nearing saturation and single-user revenue declining year after year. On the other hand, internet platforms, leveraging their content and service advantages, dominate the core value of the industrial chain, while operators gradually become mere conduits for data transmission, with their revenue from data transmission being intercepted by platform providers, leaving them with only meager channel fees.

This "conduit" dilemma is not a short-term phenomenon but an inevitable result of the mature development of the mobile internet. Operators' network infrastructure serves as the foundation of the digital economy, yet it has always been at the bottom of the value distribution chain. In this context, any opportunity to break free from the conduit role and reconstruct profit models will be highly valued by operators. The boom in AI large models and agents provides precisely such an opportunity. Tokens, as the core unit of measurement for AI computing power consumption, hold far greater value than traditional data traffic. Selling tokens essentially represents a leap from "selling connectivity" to "selling intelligence," an inevitable attempt by operators to escape the fate of being mere conduits and reshape their industry status.

Secondly, the explosive development of AI agents provides a core opportunity for operators to sell tokens. Since the beginning of this year, applications represented by various AI agents have experienced explosive growth. Especially after trends like "raising virtual pets" and "virtual horse racing" became prevalent in many industries, the demand for computing power across society is no longer an implicit backend support but an explicit frontend consumption. Tokens, as the smallest unit of measurement for large models to process information, have seen their consumption surge exponentially.

This change has brought new hope to operators: Just as users were willing to pay for "pipeline bandwidth" for data transmission in the mobile internet era, users in the intelligent economy era have a reason to pay for "computing power consumption" that produces intelligence. Operators have realized that tokens are essentially the "new traffic" of the AI era, possessing characteristics of standardization, measurability, and high-frequency transactions. By encapsulating abstract AI capabilities into standardized products similar to phone credit packages, operators have successfully demystified cutting-edge technologies, making them understandable and purchasable daily services for ordinary consumers and small and medium-sized enterprises. This opens a door for operators to access a trillion-dollar computing power consumption market.

Thirdly, the three major operators possess irreplaceable core advantages. In this transformation, the three major operators have unique strengths that other tech giants cannot match, namely their nationwide computing power network infrastructure and an extensive channel system that reaches every corner. On one hand, operators possess a nationwide backbone optical fiber network and data centers and edge computing nodes distributed across various regions. This "cloud-network integration" physical infrastructure is the fundamental guarantee for low-latency, high-reliability operation of AI inference. According to a report by China Business Journal, summarizing the annual reports of the three major operators reveals that AI-related computing power investment and revenue have become the biggest highlights for operators. In 2025, China Telecom's computing power infrastructure investment reached 20.2 billion yuan, accounting for 25% of its total annual investment; China Mobile's computing power service revenue in 2025 was 89.8 billion yuan, a year-on-year increase of 11.1%, with computing power and intelligent service revenue accounting for 20.2%; additionally, China Unicom's computing power business revenue accounted for over 15%.

On the other hand, and most crucially, operators firmly hold billions of individual mobile users and tens of millions of government-enterprise customer resources. As of the end of April this year, the total number of 5G package users among the three major telecom operators reached 1.213 billion. This user data is the most valuable resource for operators.

For the vast majority of ordinary users, the barrier to using AI often lies in complex registration processes, foreign currency payments, or obscure technical interfaces. The token packages launched by operators directly integrate with the phone credit payment system, allowing users to purchase AI computing power without additional learning costs, just like topping up their phone credit. This extreme convenience and natural trust endorsement significantly lower the barrier to using AI, enabling operators to quickly seize consumption entry points in lower-tier markets and long-tail users. This is a competitive edge that no pure internet large model company can achieve through heavy spending in the short term.

Fourthly, what are the issues with the token business? Despite the inherent advantages of operators in laying out token businesses, the potential challenges in this field cannot be ignored. Whether selling tokens can truly become a new profit opportunity remains highly uncertain. On one hand, tokens are not scarce resources; they are essentially the tangible products of computing power and energy, and they are on a continuous path of price reduction. The marginal cost of tokens is composed of real GPU computing power and electricity costs. As the user base grows and token consumption increases, costs will continue to rise without any automatic convergence, putting operators under tremendous pressure in cost control. A slight misstep could lead them into a dilemma of "increased revenue but decreased profits."

On the other hand, there are fundamental differences in the value of tokens. The same 1 million tokens, when run on a small model versus a top-tier model, carry vastly different levels of intelligence, with markedly different market recognition and pricing potential. Currently, operators are more like "middlemen" for tokens, primarily relying on external core models to provide intelligent services. If they cannot master the R&D capabilities of core models, they will struggle to build differentiated competitive advantages and may ultimately become mere channel workers for core model manufacturers, with their profit margins tightly constrained.

It can be said that although selling tokens by the three major operators appears to be a lucrative business, the underlying logic is vastly different from the mobile internet traffic that operators once controlled. Whether it can truly achieve new business growth logic remains uncertain. After all, while the token economy is promising, no one can guarantee profitability. Whether operators can break through their current revenue dilemmas with the token economy is actually a test for each telecom operator.

END