China Telecom Starts Selling 'Domestic Tokens'

Not Selling What They Have, But Delivering a Systematic Output

China Telecom has taken a pivotal step in AI commercialization — it has begun selling "domestic tokens."

Unlike the common large model API services available on the market, China Telecom's domestic tokens are not simply a packaged offering of a single large model's inference capabilities. Instead, they are built on a comprehensive "National Cloud + National Chips + National Models" AI infrastructure system, delivering full-stack domestically produced capabilities spanning from underlying chips and cloud computing platforms to upper-layer large models. This means every token generated runs on domestic chips, is hosted on a domestic cloud platform, and is powered by domestic large models.

The Integrated 'National Cloud + National Chips + National Models' Architecture

"National Cloud" refers to CTYun (China Telecom Cloud), which serves as the core foundation of China's national cloud, providing secure and controllable cloud computing infrastructure. "National Chips" signifies the adoption of domestically produced AI chips at the computing layer, reducing singular dependence on foreign high-end GPUs. "National Models" encompasses China Telecom's self-developed or deeply co-developed domestic large model ecosystem, covering a range of scenarios from general-purpose dialogue to industry-specific verticals.

The deep integration of these three pillars gives China Telecom end-to-end autonomous and controllable capabilities from chips to models. This is not merely a product launch but an innovation in AI infrastructure delivery — encapsulating a complex underlying technology stack into standardized token services, enabling government and enterprise clients and developers to use domestic AI capabilities as easily as turning on water or electricity.

Why Is a Telecom Operator Leading the Charge?

In the current AI industry landscape, internet giants and AI startups typically play the role of model providers. A telecom operator's entry may seem like a cross-industry move, but it carries unique advantages.

First, China Telecom operates a nationwide computing network and data center resources — a natural foundation for hosting large-scale AI inference workloads. Second, as a central state-owned enterprise and the primary builder of the national cloud, China Telecom holds an irreplaceable trust advantage in security compliance and autonomous controllability. This is especially significant for clients in critical sectors such as government affairs, finance, and energy, where the appeal of "domestic tokens" extends far beyond pricing.

At a deeper level, telecom operators are transitioning from their traditional role as "pipeline" providers to "computing + AI service" providers. Selling tokens is essentially packaging computing power, models, and platform capabilities into on-demand, pay-per-use AI services — a model that aligns closely with operators' expertise in large-scale service operations.

The Market Potential of Domestic Tokens

Currently, inference costs for global AI large models remain high, while Chinese enterprises using overseas model APIs also face multiple risks including cross-border data transfers and supply chain security concerns. China Telecom's domestic tokens address two core market pain points: cost controllability and security and trustworthiness.

For government and enterprise clients, using domestic tokens means data stays within national borders, computing power is guaranteed, and the supply chain is free from foreign constraints. For small and medium-sized developers, standardized token services also lower the barrier to accessing domestic AI capabilities.

From an industry trend perspective, AI is moving from a phase of "competing on models" to a new stage of "competing on infrastructure." Whoever can provide stable, secure, and cost-effective token supply will occupy a critical position in the next wave of AI application proliferation.

Outlook: Accelerating Domestication of AI Infrastructure

China Telecom's sale of domestic tokens sends a clear signal: autonomous controllability in China's AI industry is evolving from a slogan into commercially viable products. As domestic chip computing power continues to improve and domestic large model capabilities keep iterating, the cost-effectiveness and competitiveness of "domestic tokens" are expected to strengthen further.

It is foreseeable that China Mobile, China Unicom, and other state-owned cloud providers may follow suit with similar strategies, driving domestic AI infrastructure toward economies of scale. When tokens become a standardized commodity like data traffic, the foundation of China's AI industry will become significantly more robust.