Source: Shenlan Finance

At the recently concluded 2026 World Artificial Intelligence Conference (WAIC), China’s three telecom giants—China Mobile, China Telecom, and China Unicom—made a synchronized move to highlight their roles in the burgeoning ‘Token economy.’

In simpler terms, they’re positioning themselves as the essential ‘shovel sellers’ in the AI gold rush.

Behind this strategic alignment lies a Token market that’s exploding in scale. By March 2026, China’s daily average Token processing volume soared to 140 trillion, representing over a 1,000-fold increase from early 2024.

The data sends a clear message: Tokens are replacing raw computing power as the new benchmark for AI industry success.

This raises two critical questions: Who are the key players supplying these Tokens? And how will this shift redefine competitive dynamics across the AI ecosystem?

1

The Telecom Giants’ Strategic Gambits

While all three operators are betting big on Tokens, their approaches reveal distinct strategic priorities.

Let’s start with China Telecom. The earliest mover among the trio, China Telecom has articulated the clearest vision. During WAIC 2026, Chairman Ke Ruiwen defined Token operations as ‘delivering AI services to customers.’

On the production side, China Telecom isn’t satisfied with merely distributing Tokens—it aims to industrialize their creation. The multi-billion-yuan ‘Token Factory’ initiative exemplifies this ambition. Recently, its Ningxia branch launched a centralized procurement project for 2026 Token generation capacity, with an estimated value of 16.451 billion yuan.

On the service side, China Telecom unveiled the Xingchen TokenHub platform, acting as a centralized hub for Token services. Enterprises can now access diverse AI capabilities through a single account and API connection, eliminating the need for separate integrations with each model provider.

China Mobile, meanwhile, centers its Token strategy around industrial internet. At WAIC 2026, it officially launched ‘Mobile Tiangong,’ a unified industrial internet brand. Leveraging the platform’s open ecosystem, it attracts industrial software developers, smart equipment providers, and system integrators. These partners encapsulate industrial AI models, agents, and data tools into standardized services, enabling pay-per-use billing based on Token consumption.

China Unicom has adopted a ‘full-stack Token operations’ approach. It proposes an ‘Agent + Token + AI Cloud’ framework to create an end-to-end Token service system covering creation, storage, transfer, pricing, and application. Simultaneously, it focuses on cost optimization, having independently developed the Uni-Infer framework. This four-dimensional solution combines computing-network-model collaboration, model-computing fusion, model optimization, and multi-modal intelligence to slash inference costs.

The operators’ collective pivot signals a paradigm shift: Tokens have evolved from a niche technical concept to a core production factor in AI.

2

The Computing Power Model’s Third Act

The ‘Token Factory’ concept gained prominence when NVIDIA CEO Jensen Huang described modern AI data centers as ‘factories producing intelligence.’ At WAIC 2026, this vision resonated across the industry.

Token factories represent the third evolution in computing power monetization:

• First Generation: Selling hardware (GPU servers)
• Second Generation: Selling computing power (cloud services billed by usage hours)
• Third Generation: Selling Tokens (customers focus on price-per-million-Tokens and response times)

This transformation is driven by soaring Token consumption. In January 2025, China’s enterprise-level MaaS market averaged 1.6 trillion Tokens daily. By December 2025, this surged to 9.6 trillion, with projections for 2026 reaching 40,000 trillion—20 times the 2025 volume.

As AI agents replace conversational AI as the primary growth engine, Token consumption per task has jumped from tens of thousands to millions, opening vast untapped markets.

This opportunity hasn’t escaped cloud leaders like SenseTime or hardware giants like Sugon and Huawei, all now competing in the Token factory space. Meanwhile, venture-backed independent Token producers are rapidly gaining traction.

On June 30, Silicon Flow filed for a Hong Kong Stock Exchange listing under Chapter 18C, aiming to become the city’s first Token factory IPO. Prior to this, the company had raised seven funding rounds, achieving a post-money valuation of 7.74 billion yuan.

Other players are following suit. Tsinghua-affiliated Qujing Technology secured Series A funding in July 2026, raising over 1 billion yuan within six months of its 2023 founding. Infinite Chip announced another 700 million yuan in May, bringing its total funding to over 2.2 billion yuan.

3

Reshaping the AI Value Chain

The telecom operators’ synchronized move at WAIC 2026 marks the Token economy’s transition from concept to reality. It also signals China’s AI industry’s maturation—much like how energy infrastructure evolved from competitive resource extraction to standardized utility services.

Previously, companies scrambled for computing power to train proprietary models. Now, with foundation models stabilized, the focus has shifted to utilization efficiency, creating massive Token demand.

This shift benefits the entire value chain:

• Upstream: The Token model reduces reliance on NVIDIA’s hardware dominance. Domestic chipmakers like Cambricon, Hygon, Biren, and Enflame Technology can now compete on cost and system optimization. System integrators (e.g., Inspur Information, Sugon) gain prominence as essential enablers.
• Downstream: Tokens democratize AI access. SMEs can now manage operational risks through pay-as-you-go Token fees instead of risking bankruptcy on expensive hardware bets. This fuels industry-wide innovation.

However, Token selling isn’t without challenges. Operators must balance technical reliability with cost control. Effective scheduling algorithms and demand forecasting become critical—success here yields greater profits than traditional hardware sales, while failure carries proportionally higher risks.

4

Conclusion

The shift from computing power to Token sales, from hardware arms races to system efficiency, marks AI’s ‘second half’ as declared at WAIC 2026.

The telecom operators’ collective pivot is just the beginning. As Tokens become the ‘digital currency’ of AI, winners will be those who produce them most efficiently and affordably.

Ultimately, this transformation positions AI to become as ubiquitous and essential as electricity—a utility benefiting all of society.

*Disclaimer: This article reflects the author’s opinions only. Market investments involve risk; this content does not constitute personalized investment advice.

Shenlan Finance New Media Cluster originated from the Shenlan Finance Journalist Community with a 16-year legacy. As China’s leading financial new media outlet, we focus on the nation’s most valuable companies, cutting-edge industries, and emerging regional economies, delivering actionable insights for investors, executives, and the middle class. Follow us for more.