AnswersPublished41 sources
China Mobile’s Token Plan: Why AI Usage Is Becoming a New Telecom Business
China Mobile launched nationwide Token packages in May 2026, starting at RMB 5 per month, while daily Token calls in China exceeded 140 trillion by March 2026. Its MoMA platform aims to make Tokens a common, billable unit across models, computing, applications, and users through unified measurement, authentication,...
Published byEdited with GPT-5.6 LunaImages generated with GPT Image 1.5
Research answer
Key takeaways
- China Mobile launched nationwide Token packages in May 2026, starting at RMB 5 per month, while daily Token calls in China exceeded 140 trillion by March 2026.
- Its MoMA platform aims to make Tokens a common, billable unit across models, computing, applications, and users through unified measurement, authentication, billing, and settlement.
- China Telecom and China Unicom have launched competing packages, but inconsistent model costs, unclear consumption, pricing pressure, and immature standards remain major obstacles to a true national Token market.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What are China Mobile’s plans for nationally unified Token package pricing in consumer and enterprise markets, why is China’s rapidly increa. Article summary: China Mobile is trying to turn Token consumption into a national, carrier-style AI service: standardized, billable, and bundled with models, connectivity, devices, cloud PCs, and security. The opportunity is substantial,. Topic tags: general, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers, clic
China Mobile is attempting to make AI compute behave more like a telecom utility: packaged, metered, billed, and distributed through an existing national network. The company has already announced nationwide Token packages, including an entry price reported at RMB 5 per month, while its longer-term plan is to standardize pricing for both public and government-and-enterprise customers. 1
17
The timing reflects the rapid growth of AI inference. China’s average daily Token calls rose from roughly 100 billion in early 2024 to more than 140 trillion by March 2026—an increase of more than 1,000 times. 3
6 But the surge in usage does not automatically prove that consumer Token subscriptions will be profitable. The strongest near-term opportunity is likely to come from developers, enterprise applications, agents, coding, and professional content production, where AI is embedded into repeatable workflows.
1
48
What China Mobile is planning
China Mobile says it has designed multiple Token pricing tiers and tested them across several provinces. Its next step is nationally unified pricing for both public and government-and-enterprise markets, focused on high-frequency scenarios and high-value customers. The company describes the intended products as flexible combinations of Tokens, models, network traffic, terminals, and member benefits that customers can order on demand. 17
This is broader than selling a fixed allowance of model calls. China Mobile is trying to connect the consumption layer to its existing cloud, network, device, and billing businesses. In that model, a customer could buy AI capacity alongside a cloud PC, cloud phone, connectivity service, or application rather than managing each component separately. 1
17
The company’s rollout has followed a “national framework, local experimentation” pattern. Examples reported during the pilot phase include:
- Jiangsu: RMB 5 for a 2.5 million-Token trial package.
- Shanghai: RMB 1 for 400,000 Tokens on a pay-per-use basis.
- Beijing: compute-use packages from RMB 5.99 and monthly packages from RMB 24.99.
1
China Mobile announced nationwide availability in May 2026, with packages starting at RMB 5 per month and coverage for users nationwide. 5
24 Later statements about introducing nationally unified pricing indicate that nationwide availability and fully standardized national tariffs are not necessarily the same milestone. Local products and purchase channels may continue to differ while the group-level measurement and settlement system is being built.
1
17
22
Why Token usage matters to telecom operators
Traditional telecom companies are looking for ways to monetize more than connectivity. Token usage offers a potential new billing unit for AI inference: every prompt, response, tool call, or agent action can consume processing capacity, creating a measurable link between customer activity and compute demand. 19
28
That opportunity is strategically important because carriers already possess several assets AI providers need:
- established billing and payment systems;
- large customer bases and distribution channels;
- cloud and data-center infrastructure;
- network connectivity and traffic management; and
- enterprise sales relationships.
China Mobile’s stated ambition is to turn Token into a common currency connecting models, computing capacity, applications, and users. Its broader Token ecosystem plan includes unified charging, settlement, authentication, and cross-application circulation. 19
28
The practical implication is a shift from selling only network access toward selling managed AI capability. For enterprises, that could mean paying for measured inference or intelligent-computing output instead of reserving fixed infrastructure by server hour. That transition remains a strategic direction rather than a mature, proven commercial model. 1
52
MoMA is the platform behind the strategy
China Mobile’s Mobile Model Service Platform, or MoMA, is designed as a unified access layer for multiple models and computing resources. Its stated proposition is “one integration, intelligent selection, affordable access, and secure and reliable” model service. Instead of integrating separately with every model provider, a customer can use one gateway for model access, routing, authentication, metering, and settlement. 17
18
The platform is intended to solve several problems at once:
- Model access: connect customers to a broad selection of models through one integration.
- Routing: choose models or computing resources according to priorities such as cost, performance, or balance.
- Metering: convert different consumption patterns into a more consistent measurement system.
- Settlement: allocate payments among the carrier, model providers, compute resources, applications, and other partners.
- Continuity and security: provide managed access rather than leaving every customer to handle availability and authentication independently.
18
19
China Mobile says MoMA has connected more than 300 models and that its centralized Token operations reduce unit Token costs by about 30%. Those figures are company or ecosystem claims, not independently audited performance results. 12
18
The standardization problem is real. Different models can vary in tokenization, input and output pricing, context handling, reasoning modes, multimodal processing, and tool use. A nominal allowance such as 10 million Tokens therefore does not necessarily represent the same amount of usable AI work across models. China Mobile’s “unified compute measurement” approach attempts to translate those differences into standardized points or credits. 1
24
How the three major carriers compare
China Mobile, China Telecom, and China Unicom moved into Token-based pricing within a short period, but their approaches differ in emphasis. 2
4
China Mobile: national reach and cross-model standardization
China Mobile combines nationwide distribution with a platform strategy centered on MoMA. Its reported entry price is RMB 5 per month, and its products have been linked with cloud PCs, cloud phones, and AI applications. Its provincial pilots have included pay-per-use and monthly formats, giving the company room to test how customers understand and consume AI capacity. 1
5
24
The differentiator is not simply the lowest advertised price. China Mobile is trying to establish a common measurement, access, and settlement layer across multiple models and applications. 17
18
China Telecom: clearer group-level consumer and business tiers
China Telecom launched trial-commercial Token packages on May 17, 2026. Its individual and family plans were reported at RMB 9.9, RMB 29.9, and RMB 49.9 per month for 10 million, 40 million, and 80 million Tokens, respectively. Developer and small-business packages were reported from RMB 39.9 to RMB 299.9 per month, with larger Token allowances and optional connectivity or security services. 7
9
32
That structure resembles a conventional mobile-data tariff: distinct consumer and professional tiers, fixed monthly allowances, and add-on services. It may be easier to explain than a fully dynamic compute bill, although it still leaves customers needing to understand what each Token allowance buys in practice. 11
54
China Unicom: developers, teams, and one-person companies
Unicom Cloud’s personal Token Plan was reported to start at RMB 15 per month for 6 million Tokens, with higher personal tiers at RMB 30 and RMB 45. Its team plans use Credits rather than raw Tokens and were reported at RMB 198, RMB 698, and RMB 1,398 per month. 33
36
44
Shanghai Unicom also targeted one-person companies, or OPCs, with a temporary 30-million-Token testing allowance and discounts on selected plans. Other regional offerings combined Tokens with coding tools, AI cloud desktops, communications, broadband, and agent services. 32
33
42
The comparison shows that the market is not yet using one universal commercial language. One carrier may quote Tokens, another Credits, and a third may combine AI usage with a specific application or cloud service. That variety may help carriers test demand, but it complicates direct price comparisons. 33
44
54
Why enterprise demand is more important than consumer adoption
For ordinary consumers, Token pricing has a basic communication problem. People understand what a gigabyte of mobile data roughly represents because they can associate it with browsing, video, or downloads. They generally cannot predict how many useful conversations, documents, images, coding tasks, or agent actions a given Token allowance will support. The result is that a cheap-looking package may still feel expensive if the balance disappears quickly. 1
55
Consumers also often have access to free or subsidized AI chat services, especially for casual conversation, search, and text editing. That reduces the immediate reason to purchase a separate Token allowance. 1
48
Enterprise customers have a clearer use case. A company may need predictable AI capacity for customer service, software development, document processing, content generation, internal assistants, or agent workflows. In those settings, Token consumption can be tied to an application, a team, a service-level requirement, or a business process rather than sold as an abstract allowance. 1
17
That suggests a likely division of labor:
- Consumer packages help carriers distribute AI services, attach usage to devices and cloud products, and build user habits.
- Enterprise packages are more likely to generate recurring, measurable demand because companies call models through APIs and embedded applications.
The consumer market could become more significant if AI phones, household devices, and autonomous agents begin generating continuous model activity. For now, available reporting supports a more cautious conclusion: consumer Token packages are an important strategic experiment, while enterprise usage is the more credible early revenue pool. 48
50
52
The obstacles to truly unified national pricing
1. Tokens are difficult to explain
Token is a technical unit, not a direct measure of a completed task. Consumption depends on the model, context length, input-output ratio, reasoning mode, tools, and the type of data being processed. Even a simple request can produce unexpectedly high usage if the connected application adds hidden context or tool calls. 1
35
A viable retail product therefore needs more than a large quota. It needs clear usage dashboards, task-level estimates, warnings, spending controls, and explanations of how different models consume the allowance.
2. A common unit does not automatically create comparable value
Model providers use different pricing and accounting methods. Some distinguish input, output, caching, or model tiers; others package usage inside an application or subscription. Multimodal and agent workloads make the comparison harder still. 34
42
54
China Mobile’s standardized-credit layer could make the customer experience simpler, but it must be supported by transparent conversion rules. Otherwise, customers may see a common label while receiving very different amounts of useful work from different models.
3. National settlement is harder than provincial promotion
A nationwide platform must track the cost and revenue associated with compute pools, models, applications, channels, provincial operating companies, enterprise accounts, and service guarantees. It also needs consistent authentication, usage records, billing rules, and partner settlement. China Mobile has identified these capabilities as core parts of its unified platform strategy, but building them at scale is materially more difficult than launching a discounted regional bundle. 18
19
4. Carrier prices face direct competition from model providers
The value proposition can weaken quickly when model companies cut their API prices. Reporting has already raised concerns that some carrier packages may be more expensive than direct model-provider pricing on a per-million-Token basis, while their bundled services and cross-model access may not always be transparent enough to explain the difference. 34
53
Carriers will therefore need to compete on more than raw Token price. Their advantages may instead be reliability, security, billing convenience, model choice, enterprise support, network integration, and the ability to combine compute with cloud and connectivity services.
5. The business model is not settled
Customers may ultimately pay for raw Tokens, model capability, completed agent tasks, application subscriptions, or business outcomes. A low monthly Token package is easy to market, but it may not reflect the cost of intensive reasoning, long context, multimodal processing, or tool use. 1
52
The more AI becomes embedded in business workflows, the less useful a simple Token quota may become. Enterprises may prefer budgets tied to tasks, service levels, or applications rather than a technical unit that does not map neatly to value.
6. The shift from server hours to Token billing requires operational discipline
Token-based compute rental could align payment more closely with actual usage than fixed server-hour contracts. But it also requires accurate metering, capacity planning, quality-of-service guarantees, fraud controls, security protections, and clear responsibility when a model is slow, unavailable, or produces poor results. 1
20
In other words, Token billing is not merely a new line on a telecom invoice. It is a new operating model spanning infrastructure, model providers, applications, networks, and customer support.
What to watch next
The next meaningful test is not whether carriers can advertise a low entry price. It is whether they can make Token consumption predictable and valuable enough for customers to renew.
Three indicators matter most:
- Enterprise renewal and workload depth: whether businesses move from trials to recurring production use.
- Transparency of measurement: whether customers can compare models, tasks, and effective costs without learning each provider’s accounting system.
- Integration beyond the quota: whether Token packages become useful components of cloud, security, devices, agents, and managed AI services rather than isolated prepaid balances.
China Mobile’s national strategy is therefore best understood as an infrastructure and distribution bet. The company wants to become a managed AI-usage platform, not simply another seller of model credits. Its May 2026 nationwide packages mark an important commercialization step, but the evidence still points to an early market in which pricing, standards, consumer demand, and enterprise economics remain unsettled. 1
17
19
The central question is no longer whether Tokens can be counted. It is whether carriers can turn that count into a transparent unit of value that customers trust—and whether the resulting business is strong enough to move telecom operators from selling connectivity toward selling intelligent services.
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "China Mobile’s Token Plan: Why AI Usage Is Becoming a New Telecom Business"?
China Mobile launched nationwide Token packages in May 2026, starting at RMB 5 per month, while daily Token calls in China exceeded 140 trillion by March 2026.
What are the key points to validate first?
China Mobile launched nationwide Token packages in May 2026, starting at RMB 5 per month, while daily Token calls in China exceeded 140 trillion by March 2026. Its MoMA platform aims to make Tokens a common, billable unit across models, computing, applications, and users through unified measurement, authentication, billing, and settlement.
What should I do next in practice?
China Telecom and China Unicom have launched competing packages, but inconsistent model costs, unclear consumption, pricing pressure, and immature standards remain major obstacles to a true national Token market.