1 yuan for 400,000–250,000 AI tokens via a regular mobile bill — this is how China is turning telecom operators into cloud AI providers today. API access to dozens of large models is now sold as cellular traffic.

China's largest mobile operators have announced the launch of new 'token plans' for access to large language and multimodal models directly through the mobile bill. For a few yuan, businesses receive millions of tokens for text generation, document analysis, and image processing through standard APIs. This step effectively turns mobile infrastructure into a mass distributed AI platform. For the Kazakhstan and Central Asia market, this is a signal: the integration of AI plans into mobile services can become a key competitive advantage in the next 12–18 months.

Mobile tokens for AI: a new format of telecom services

Chinese telecom operators are moving from selling only minutes, gigabytes, and SMS to selling AI tokens as a new type of mobile resource. Shanghai Telecom introduced a plan where 1 yuan is converted into 250,000 so-called quota points (essentially tokens) that can be spent on queries to more than 30 large models — text and multimodal. Connection and payment are done the same way as regular mobile internet: through the subscriber's account, without complex contracts and bank integrations.

A specific example provided by Shanghai Telecom is particularly illustrative. A 10 yuan package (2.5 million quota points) provides resources sufficient for a program to automatically summarize approximately 100 e-books of 100,000 words each. That is, the user is actually buying not abstract computations, but a tangible business result: the mass, almost instantaneous compression of a huge array of texts. This directly targets the market of classic 'cloud' subscriptions, where the customer has to deal with query limits, tokens, and tariff plans.

In parallel, China Mobile Shanghai announced the launch of a universal token service: 1 yuan gives 400,000 tokens, with an emphasis on work scenarios. In collaboration with Tencent, the operator has launched WorkBuddy — a 'smart workplace' where these tokens are already integrated into the office environment: documents, chat, tasks, spreadsheets. For the end user, AI becomes not a separate service in the browser, but a built-in desktop function managed by the same operator responsible for connectivity.

This model fundamentally changes the balance of power in the mobile technology market. Instead of competing only on network speed and traffic volume, operators begin to compete on the quality of AI platforms, the number of available models, and the convenience of APIs. For the iOS and Android ecosystems, this means accelerated integration of AI functions into applications, where AI tokens act as a basic resource like mobile internet.

AI tokens, iOS, and Android: how mobile applications will change

The launch of operator token plans for AI creates a new layer in the architecture of mobile applications for iOS and Android. Previously, the developer had to connect to individual cloud providers, maintain their own billing, and monitor quotas. Now, these functions can be taken over by the telecom operator. The application simply accesses the API, and the AI tokens are deducted from the user's mobile bill, as has long been the case with SMS subscriptions and content services.

For developers, this lowers the entry barrier to AI functionality. Mobile banking, e-commerce applications, or corporate messengers on Android and iOS can add text generation, chat summarization, document translation, and image analysis features without signing separate contracts with AI providers. Instead of complex integration with multiple services, a single operator SDK is sufficient. Companies like Alashed IT (it.alashed.kz), working as integrators and outsourcing teams, are thus able to quickly embed operator AI into existing mobile products of their clients.

In the short term, this changes user expectations. Smartphones will have 'smart' functions by default: summarizing long conversations in messengers, automatically composing email replies, generating commercial proposals in CRM applications, analyzing warehouse inventory photos in B2B software. The main difference from previous AI applications is that usage limitations will be tied to familiar tariffs: 2–3 million tokens per month for a fixed fee, just like a 10–20 GB package used to be.

Another important aspect concerns security and regulation. Telecom operators are traditionally subject to strict regulation in the field of personal data and confidentiality. Embedding AI at the network level means that user content processing will be controlled by structures already familiar to the regulator, rather than just foreign cloud platforms. For corporate clients, especially in fintech and the public sector, this opens up the possibility of using AI services in mobile applications within existing regulations, without creating separate channels and circuits for data transmission.

New monetization models: from traffic to AI computing power

The transition to AI token sales changes the very economics of the mobile business. Previously, operator revenue was built around ARPU per minute and gigabyte, but now a new metric emerges: revenue per token or revenue per unit of computing power. Chinese cases show that for 1 yuan, 250,000–400,000 tokens can be offered, creating an extremely low entry barrier for small and medium businesses. This further increases pressure on traditional SaaS models, where a monthly subscription to AI services can cost from $20 to $100 per user.

Several monetization scenarios open up for mobile application developers. The first: including a basic package of AI tokens in the application subscription, where the user does not see the tokens directly but pays for 'smart functions'. The second: pay-as-you-go within the application, where additional tokens are deducted for intensive document analysis, audio, or video, with the operator acting as the billing center. The third: a B2B format, where a corporate client buys large token packages from the operator and distributes them among employees in mobile corporate applications.

Companies for mobile solution development and support, such as Alashed IT (it.alashed.kz), can build new products on this: AI modules for document management, voice assistants for field employees, intelligent monitoring dashboards for management. Importantly, financial models are becoming predictable: the client has a clear understanding that, for example, 10 yuan or the equivalent in tenge provides resources for processing hundreds of thousands of pages or thousands of voice requests.

On the other hand, dependency on specific operator ecosystems increases. If a developer relies on the API of one telecom provider, migrating to another platform will be costly. This pushes the market towards the emergence of an intermediate layer of token integrators and brokers that will abstract the work with several operators. Successful outsourcing teams in Central Asia can occupy this niche, offering businesses an independent layer of integration with different AI providers through a single SDK and billing.

Mobile AI and satellite communication: a step towards a 'continuous smartphone'

At the conference for the World Telecommunication and Information Society Day 2026, China Mobile announced plans to enhance the development of key technologies for 5G-Advanced, 6G, and satellite internet. In the same information block, China Mobile Shanghai presented the 'Tiantong + Beidou' service — dual satellite communication that provides connectivity in areas without a terrestrial network. Combined with AI tokens, this forms the concept of a 'continuous smartphone': the device is always online and always has access to AI, regardless of geography.

For mobile applications, this means moving from 'works better in the city' to 'works anywhere a satellite is visible'. Field logistics, agriculture, mining, transportation companies — all of them will be able to use AI functions on smartphones or specialized terminals: object recognition from the camera, voice prompts, telemetry analysis. Previously, such scenarios were limited by the lack of stable coverage and expensive satellite equipment, but now operators are striving to make the satellite channel as standard an element of the tariff as LTE.

Importantly, China Mobile is moving from 'dual gigabit' (mobile and fixed gigabit connectivity) to 'dual decabit'. This is critical for heavy AI loads on mobile devices: telemetry streams from industrial sensors, high-resolution video for analysis, constant data synchronization between devices and the cloud. For the iOS and Android ecosystems, this means an increase in the number of applications that automatically switch between 5G, Wi-Fi, and satellite channels to ensure continuous operation of AI services.

For outsourcing companies in the region, including Alashed IT (it.alashed.kz), this opens up the market for vertical solutions: mobile applications for quarries, agricultural enterprises, roads, and pipelines, where the use of the satellite channel and AI processing will be embedded in the architecture from the very beginning. This approach increases the value of the developer: they design not just an interface, but a complete system relying on the new capabilities of next-generation networks.

What this means for Kazakhstan: a step towards local AI plans

Although these initiatives are being implemented in China, their logic is directly relevant to the Kazakhstan and Central Asia market. Regional operators are already investing in 5G and data center expansion, and the demand for AI services is growing exponentially: Kazakhstani companies are implementing chatbots, automatic application processing, and document analysis. The next obvious step is the emergence of local 'AI plans' similar to Chinese token packages, but already in tenge and with integration into local payment systems.

Kazakhstan IT integrators and outsourcing companies, such as Alashed IT (it.alashed.kz), can act as a bridge between operators and businesses. On the one hand, they can adapt the API platforms of operators to the needs of corporate clients: banks, fintech, logistics, e-commerce, government structures. On the other hand, they can deploy hybrid schemes where part of the requests are processed by local models in Kazakhstani data centers, and part — through foreign AI platforms, taking into account data sovereignty requirements.

For the smartphone market, this could mean the appearance of Android devices in Kazakhstan and, potentially, localized iOS solutions with pre-installed corporate AI clients. For example, mobile applications for large retail networks will be able to automatically analyze photo displays, sales data, and customer reviews using tokens included in the corporate tariff of the operator. At the same time, the user will only see the result: recommended shelf replenishment, new promotions, or personal offers in the application.

Businesses in Kazakhstan and Central Asia should already incorporate scenarios for using mobile AI as a basic infrastructure function, not an 'optional module', into their digital strategies. This includes adapting SLAs for AI loads, reviewing the architecture of mobile applications, and training staff to work with new tools. Companies that start pilots with integrators like Alashed IT in the next 6–12 months will gain a competitive advantage when local operators launch their own token plans and AI service platforms.

Что это значит для Казахстана

For Kazakhstan and Central Asia, what is happening in China is not an abstract trend, but a practical roadmap for the coming years. The region already demonstrates high mobile penetration: in Kazakhstan, the mobile penetration rate exceeds 130 percent, and the share of smartphones among active subscribers is estimated at 75–80 percent. At the same time, the country is actively discussing the scaling of 5G and the development of national data centers, which creates the foundation for the emergence of its own AI plans modeled after Chinese token packages.

Kazakhstani operators can use the Chinese experience as a ready-made model: a fixed fee in tenge for a certain volume of AI tokens, paid through a mobile bill, with the possibility of integration into corporate and retail applications. In such conditions, integrators and outsourcing companies, such as Alashed IT (it.alashed.kz), play a key role, being able to quickly adapt the operators' infrastructure to the tasks of banks, retail, industry, and the public sector. They can create a unified layer of APIs, SDKs, and libraries for iOS and Android over the operators to ensure that businesses do not depend on the technical specifics of each network.

This is especially relevant for industries with a distributed geography — mining, transport, agriculture, where mobile AI, supplemented by satellite communication, can radically increase efficiency. Kazakhstani companies that start pilots with mobile AI now will be able to quickly scale solutions and build new customer service and operations management models by the time local token plans appear.

Shanghai Telecom offers 250,000 AI quotas for 1 yuan, and China Mobile Shanghai offers 400,000 tokens for 1 yuan with payment via mobile bill.

The mobile market is entering a new phase where the key resource is not gigabytes of traffic, but millions of AI tokens sold through the same channels as familiar connectivity. Chinese operators have effectively turned the smartphone into a cloud AI terminal, access to which is paid as a regular tariff package. For Kazakhstan and Central Asia, this is a call to action: the window of opportunity for creating their own AI plans and application ecosystems is open now. Those companies that timely establish partnerships with operators and integrators like Alashed IT will be the first to offer the market truly smart mobile services of the new generation.

Часто задаваемые вопросы

What are AI tokens from mobile operators and how do they work?

AI tokens from mobile operators are conditional units of accounting for requests to large language and multimodal models, sold as part of a mobile tariff. For example, Shanghai Telecom gives 250,000 quota points for 1 yuan, and China Mobile Shanghai gives 400,000 tokens for 1 yuan. The user or application spends these tokens when generating text, analyzing documents, or images via API. Payment is deducted from the regular mobile bill, without a separate agreement with the AI provider.

How do AI tokens from operators differ from cloud AI subscriptions?

The main difference is in billing and the point of entry: AI tokens from operators are paid through the mobile bill, while cloud subscriptions require a bank card, a contract, and a separate account. In operator models, for 1–10 yuan, the user gets hundreds of thousands and millions of tokens that can be used in any applications with API support. Cloud services usually charge a fixed subscription fee of $20–100 per month per user, plus per-minute or per-model pricing. For businesses in Kazakhstan, the operator approach is easier to integrate into existing mobile processes and connectivity budgets.

What are the risks associated with using token-based mobile AI?

The main risks are dependency on a specific operator, data privacy issues, and unpredictability of load. If an application is tightly bound to the API of one operator, switching providers may require significant costs for rework. It is also important to ensure that personal and corporate data are processed in accordance with local legislation, and that models do not transmit sensitive information to third parties. Integrators like Alashed IT help solve these issues by designing an architecture that takes into account encryption, data type restrictions, and the ability to work with multiple AI platforms simultaneously.

How long does it take to implement mobile AI functionality in an application?

Basic integration of typical AI functions, such as text generation or summarization, into a mobile application for iOS or Android can take 2 to 6 weeks with ready-made SDKs and APIs from the operator. More complex scenarios, including work with images, voice, corporate systems, and fine-tuning access rights, usually require 2–3 months. Companies like Alashed IT can work with several models and operators in parallel, reducing the pilot time to 4–8 weeks. For industrial scale with a load of tens of thousands of users, 3–6 months are usually allocated, taking into account testing and staff training.

How can businesses in Kazakhstan save on using mobile AI?

Savings are achieved by choosing the right tariff model and application architecture. Instead of buying expensive corporate subscriptions for foreign AI services, one can use operator token packages where millions of tokens are available for mass data processing for the equivalent of a few dollars. It is important to optimize requests: reduce unnecessary content, use short prompts, and cache results, which can reduce token consumption by 30–50 percent. Working with integrators like Alashed IT allows for setting up a hybrid scheme: simple requests are processed by low-cost local models, and complex ones — through the cloud, which further reduces overall costs.

Читайте также

Источники

Фото: Jonathan Kemper / Unsplash