ZTE unveils industry's first SPN computing power dedicated line CPE with built-in AI inference

• Industry's first network-computing convergence solution based on "dedicated-line access + edge inference"
• Supports fine-granularity slicing dedicated lines and edge AI inference, accommodates models with up to 10 billion parameters, and enables quick AI application development through "building block" operators without programming

Guangzhou, China, 18 November 2024 – ZTE Corporation (0763.HK / 000063.SZ), a global leading provider of integrated information and communication technology solutions, today announced the launch of its ZXCTN 810C, the industry's first SPN computing power dedicated line CPE product with built-in AI inference.

This product introduces the industry's first network-computing convergence solution based on "private-line access + edge inference", enabling rapid deployment of edge inference through operators' dedicated-line products. The device supports both fine-granularity slicing dedicated lines and edge AI inference, accommodates models with up to 10 billion parameters, and allows for quick construction of AI applications through "building block" algorithm component orchestration without the need for programming.

This device features a variety of input interfaces. In addition to 10GE fine-granularity and GE/FE network-side interfaces, it also supports industrial control interfaces such as RS10/232 and PLC, which helps operators fully leverage network coverage advantages and enhance computing through the network to facilitate the digital and intelligent upgrade of small and medium-sized enterprises.

As the digital economy booms, computing power technologies are not only evolving towards large-scale and centralized solutions, but also extending to the edge to meet the demands for real-time processing and privacy protection. Edge inference devices process data closer to the source, shortening the distance for data transmission and thereby minimizing network latency. Since local data remains in the campus network, user data security is improved. Additionally, the reduction in data volume sent to the cloud saves network bandwidth. These advantages have made this solution widely preferred.

The SPN-based edge inference offers distinct advantages. According to existing network data, China Mobile has deployed over 400,000 SPN devices on a large scale, which have been running stably for five years, providing extensive service coverage and fast O&M response. Working with Jiutian AI platform, they can also train AI model and invoke remote data for cloud-network-edge coordination and cross-campus factory data edge-to-edge coordination.

The newly released SPN CPE is a high-quality SPN access device that offers local inference capabilities through its built-in AI computing power to deploy edge inference with optimal costs. It has the following innovative advantages:

Computing-network convergence: This is the first solution to integrate slicing dedicated lines with edge inference. The slicing dedicated lines deliver high-quality network service and security isolation to computing-power applications. It is used for not only cloud-edge computing-power coordination, but also edge-to-edge coordination of cross-campus factory data. To simplify the management and control of computing network services, it innovates the "computing power + transport power" integrated management and control to schedule network and computing resources in a unified manner.

Fast deployment: It generates AI models in hours by dragging and dropping "building block" algorithm components. Various built-in algorithm components support efficient migration and deployment in different application scenarios.

Cost reduction and efficiency improvement: For instance, in a pilot project for industrial quality inspection using AI computing power dedicated lines at an electronic factory in Jiangsu, AI inference speed is up by 3~5 times compared to traditional methods, and quality inspection accuracy is improved by 75%, while operational costs are down by nearly 30%.

Diversified scenarios: A range of input interfaces are available. In addition to 10GE fine-granularity and GE/FE network-side interfaces, it supports RS10/232 and USB interfaces commonly used in industrial production, and such image, video, and audio interfaces as HDMI and AUX to easily deal with data collection, remote control, and industrial control scenarios.

Low-carbon computing power and edge intelligence are increasingly becoming key drivers of innovation and development across industries. The SPN computing power dedicated line CPE with built-in AI inference is an innovative product developed in this context. Beyond industrial quality inspection, it will be widely applied in sectors such as security, transportation, and mining. In over 50 scenarios, including face recognition, self-driving assistance, and environmental monitoring, it will leverage the advantages of computing-network convergence.

ZTE remains committed to collaborating with industry partners to drive technological innovation and create a new digital future.