Chuling Invited to Shanghai Telecom Innovation Culture Festival: AI Implementation Requires a Well-Planned Network Foundation
Recently, the "Visiting Central Enterprises in Shanghai" forum series, part of the Innovation Culture Festival, was held in Shanghai. The event was organized by the Shanghai Small and Medium-sized Enterprise Development Service Center (Shanghai SME Listing Promotion Center) and the Shanghai Technology Exchange, with China Telecom Shanghai as the organizer.
This forum directly addressed the theme of "achieving enterprise full-stack deployment needs through AI + integrated software and hardware capabilities." It brought together business development leaders for government and corporate clients from various districts, frontline account managers, enterprise representatives, and ecosystem partners to engage in in-depth discussions on enterprise AI implementation. Intellifusion, a deeply involved ecosystem partner, was invited to attend and participate in the roundtable dialogue.
01 Industry Insights
Jiang Zhiwen, Deputy Director of the Shanghai SME Development Service Center, summarized the real challenges facing many small and medium-sized enterprises with "three no's": they don't know how to use AI, are afraid to use it, or can't afford it. These issues stem from high computing power thresholds, data security concerns, steep implementation costs, and a lack of tailored solutions for specific business scenarios.
He stated that Shanghai Telecom, a key builder of the city's digital infrastructure, has launched an AI public computing power service platform to support over 700 enterprises in their AI-driven transformation. The Municipal SME Development Service Center will continue to play its connecting role, bridging the "last mile" of AI adoption.
Xu Liang, General Manager of the Business Customer Department at China Telecom Shanghai Branch, shared frontline insights: In the AI era, digital transformation for SMEs has shifted from an "option" to a "must." Enterprise needs have evolved from single-point tools to full-stack integration spanning computing power, data, applications, and talent. However, concerns remain around security control and cost investment. "How to truly bring AI technology down to SMEs is the defining challenge for local telecom operators today."
Building on these real-world topics, other guests shared their observations based on hands-on experience, offering fresh insights for businesses and the industry.
02 Roundtable
Drawing on these frontline pain points and insights, Fang Chao, a solutions expert from Chu Ling Shares, joined hardware vendors, algorithm R&D partners, and telecom account managers for an interactive session. The panel directly addressed real-world concerns raised by enterprise representatives on-site to exchange practical ideas.
Q:After deploying AI applications, enterprises often face video analysis lag, high inference latency, and heavy data transmission loads. How should underlying capabilities be adapted?
A：AI applications place significantly different demands on underlying connectivity, compute power, and data processing compared to traditional office systems. While legacy office systems rely on intermittent, low-concurrency interactions, AI workloads typically require continuous high-volume data transfer and real-time computation. Without proactive planning of the underlying infrastructure, upper-layer applications are prone to lag, latency, and instability.
First, AI workloads require a dedicated and stable network channel.For scenarios such as video analytics, machine vision, and real-time recognition—characterized by high data volumes and continuous processing needs—it is essential to provide AI workloads with dedicated transmission channels and quality assurance. This prevents resource contention with standard office traffic. Network jitter can cause inference results to degrade or fail; therefore, deterministic connectivity is integral to AI capability itself.
Second, edge computing must be deployed closer to the source.Offload part of the reasoning capability to parks, workshops, or local nodes. Data doesn't need to be fully sent back to the cloud, reducing transmission load and significantly improving response speed.
Third, data collection must be planned in coordination with AI applications.AI performance depends on data quality. Fragmented collection and inconsistent standards will impact model training and deployment downstream. Therefore, connectivity, collection, inference, and application must be addressed as a unified system.
From Roundtable to Practice
As a vendor deeply rooted in enterprise digital intelligence, Chuling has been committed to solidifying its foundational capabilities to ensure that upper-layer AI applications not only run but also run reliably.
When it comes to concrete solutions,Chuling leverages cutting-edge technologies—SEN (Enterprise Network Agent), SPN-FTTO (Sliced Packet Networking for Fiber to the Office), and DPI (Deep Packet Inspection)—to build a proprietary, fully controllable all-optical intelligent computing network tailored for enterprises.
On one handEstablishes a deterministic full-optical channel for AI workloads, physically isolating high-continuity traffic like video analytics and machine vision from office and production networks to ensure stable inference without jitter.
On the other handECPS compute scheduling deploys inference capabilities to local nodes such as parks and workshops, reducing data backhaul and end-to-end latency by enabling AI processing at the source of data generation.
simultaneouslyLeveraging AI-driven AIOps to continuously monitor network status, link quality, and user experience—and proactively alert on issues—shifts operations from reactive troubleshooting to proactive prevention.
The "last mile" of AI deployment isn't just about installing a model. First, build a solid network foundation so applications can truly take root. Chuleng looks forward to partnering with more organizations to plant the seeds of AI in real business scenarios across every enterprise.
