en.Wedoany.com Reported - The Intelligent IP Wide Area Network (AI WAN) is becoming a key foundation for the interconnection and efficient scheduling of computing resources. In April this year, the Political Bureau of the CPC Central Committee included the computing power network in the "six networks" construction system, sparking increased discussion in the industry and mainstream media. The "Intelligent IP Wide Area Network (AI WAN) Application Promotion Action and China Tour" initiative, jointly launched by the China Academy of Information and Communications Technology and telecom operators, was recently reported by CCTV. The report noted that the intelligent computing power network, akin to a power grid dispatch center, can uniformly schedule dispersed computing resources. In cross-region deployment scenarios spanning hundreds of kilometers, computing performance loss can be controlled within 5%. This network has evolved from a "data transmission" pipeline into an intelligent foundation capable of sensing AI services, implementing intelligent scheduling and differentiated assurance, and possessing automatic risk identification and handling capabilities.

Against the backdrop of the nation's continuous push to build a national integrated computing power network, policies such as the "Action Plan for High-Quality Development of Computing Infrastructure" and the "Opinions on Deeply Implementing the 'Artificial Intelligence+' Action" have been successively introduced, explicitly requiring the promotion of cross-regional scheduling of computing resources, the construction of integrated computing-network infrastructure, and the enhancement of computing network transmission capabilities. Behind these policies lie the real needs of industrial development. On one hand, new intelligent computing scenarios drive efficient data flow between enterprises and intelligent computing centers, giving rise to new models such as "data into computing," "storage-compute separation," "remote training," and "distributed inference." As large model training scales expand, individual intelligent computing centers, constrained by space and power, need to connect computing nodes distributed across different regions to form "multi-point co-computing," enabling collaborative training and inference across intelligent computing centers. On the other hand, the rise of new terminals like AI phones, AI PCs, companion robots, and innovative applications such as cloud e-sports, industrial large models, and intelligent manufacturing robots imposes higher demands on real-time network interaction. Traditional WANs primarily serve internet services like video and web pages, addressing the issue of "whether data can be delivered," while AI WAN tackles the problems of "whether computing power can efficiently collaborate, whether data can flow securely, and whether business experience can be guaranteed."

Since the industry proposed the AI WAN direction in 2025, the technology has moved from exploration to practice. At the third "IPv6 Technology Application Innovation Competition," the AI WAN special session showcased benchmark cases from operators, enterprises, and industry partners across the country, covering scenarios such as government-enterprise intelligent computing training and inference, industry data circulation, and public user experience assurance. For government and enterprise sectors, companies like Beijing Telecom, Zhejiang Telecom, and Hebei Unicom have leveraged AI WAN integrating IPv6+, wide-area RDMA lossless transmission, SRv6, and AI intelligent scheduling technologies to solve issues like link packet loss, load imbalance, and latency fluctuations in long-distance computing collaboration. Practices show that Beijing Telecom extended hospital training and inference tasks to an intelligent computing center 240 kilometers away, achieving a computing efficiency loss of less than 5% while keeping data within the domain; Zhejiang's AI WAN lossless network supports distributed training and cloud-edge collaborative inference, with overall computing efficiency exceeding 95%; the Beijing-Tianjin-Hebei region has created a "computing corridor" for unified cross-regional scheduling of computing resources. For industry sectors, Tianjin Mobile adopted a "network-computing-data-security" four-dimensional collaborative AI WAN data circulation network architecture, establishing elastic, deterministic, secure, and trustworthy channels for cross-regional data flow. In the automotive consumption sector, this solution improved federated learning model accuracy and cooperative car company lead conversion rates by 34%; in the financial services sector, it increased enterprise compliance audit efficiency by 70%. For the public, operators like Guangdong Unicom and Shaanxi Mobile have introduced technologies such as SRv6, AI service identification, and intelligent scheduling to build an "end-network-cloud-computing" collaborative system, creating dedicated cloud access channels for services like cloud computers, cloud gaming, and cloud fitness, achieving a "zero-lag" user experience.

These practices demonstrate that AI WAN is breaking through key bottlenecks in the large-scale development of artificial intelligence: it significantly lowers the barrier for enterprises to access computing power, pushing "computing as a service" toward reality; while unlocking the value of computing power, it also balances data security and efficient factor circulation; and it continuously stimulates innovation in new applications from industrial large models and intelligent manufacturing to cloud gaming and digital humans. With the launch of the "AI WAN Application Promotion Action," the technology is moving from "point innovations" to "large-scale applications," and collaboration across the industry chain will be further strengthened. An intelligent IP WAN that "understands computing power, understands services, and understands scheduling" is gradually becoming the new foundation of the intelligent era.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com