At the opening session of the 2026 China Internet Conference, an executive from China Telecom Corporation Limited delivered a keynote address titled "Embracing the New Intelligent Era, Co-building a New Digital-Intelligent Ecosystem."
The executive stated that artificial intelligence is no longer merely a supplementary tool but has become a core force reshaping production relationships. AI's adoption is increasing across industries such as manufacturing, government services, and finance, rapidly integrating into all sectors and households, thereby accelerating the transformation of the industrial landscape. Concurrently, AI development is evolving from isolated technological breakthroughs toward a comprehensive system change characterized by the deep integration of computing power, models, data, and applications. Traditional industry divisions and value distribution models are being replaced by a new operational paradigm centered on tokens.
He emphasized that in response to this comprehensive and historic shift, China Telecom will proactively embrace and fully integrate into the trend. The company is committed to continuously enhancing its integrated system encompassing computing power, platforms, data, large models, and intelligent agents, driving the transformation of its business operations from traditional traffic management to a token-based operational model.
Building a Native AI Computing Foundation
The first pillar involves solidifying a native AI computing infrastructure. Leveraging its nationwide integrated cloud-network backbone resources, the company has established a carrier-grade, ubiquitous computing power network, enabling unified scheduling of general-purpose, intelligent, and supercomputing resources across the entire network. It has deployed three major 10,000-card intelligent computing clusters in the Beijing-Tianjin-Hebei region, the Yangtze River Delta, and the Guangdong-Hong Kong-Macao Greater Bay Area. By the end of 2025, the company plans to launch version 2.0 of its Xiran Computing Power Internet scheduling platform, with its total proprietary and accessed intelligent computing power expected to exceed 91 EFLOPS.
Strengthening Core Capabilities in Models and Data
The second focus is on fortifying core competencies in models and data. The company is developing a comprehensive, fully-domestic large model system covering all modalities and sizes. It has launched the Xingchen TokenHub operational service platform, compatible with over 100 mainstream large models. Additionally, it is building a trusted data circulation infrastructure integrating central and local systems, creating a massive, integrated data resource system.
Enabling Large-Scale Industry Adoption
The third strategy aims to empower large-scale implementation across diverse industries. Focusing on sectors like manufacturing, finance, healthcare, and modern services, the company has developed over 110 industry-specific large models and more than 350 industry intelligent agents, serving 37,000 government and enterprise clients. For office scenarios, it introduced the Xingchen Super Intelligent Agent, supporting out-of-the-box use and the accumulation of enterprise-specific capabilities. Commercially available token packages have been released to the market, significantly reducing corporate costs for AI computing power and accelerating the widespread adoption of the token-based operational model. "Our goal is to make AI easy to use, effective, and affordable, allowing the benefits of intelligent development to be shared across society," the executive stated.
Balancing Security and Industrial Development
The fourth area involves coordinating security with industrial growth. To address network security analysis scenarios, the company has independently developed the Jianwei security-specific large model and created a security product portfolio, including Cloud Pulse, to provide comprehensive protection for the entire lifecycle security of token operations.
Comments