5G-A and AI as Dual Engines: Building a Future-Oriented Foundation for Industrial Digital Intelligence

AI is leading a new round of technological revolution and industrial transformation, injecting strong momentum for innovation across industries, reshaping production methods, restructuring productivity, and accelerating the industry's transformation towards a more digitalized, connected, intelligent, and sustainable future. With the deep penetration of AI applications such as industry-specific large models, AI agents, and embodied robots into the entire production process, real-time information exchange, data processing, and decision-making increasingly require a robust digital intelligence foundation.

Challenges Faced by Industrial Digital Transformation

Due to industry differences, diverse application scenarios, and varying stakeholder interests, the infrastructure of vertical industries has long been characterized by a coexistence of multiple technologies and uneven levels of development, facing the following challenges:

• Closed architecture: Traditional industrial networks are based on the closed ISA-95 architecture, forming vertically segmented hierarchies and data silos where resources and data cannot be fully shared or circulated.
• Limited flexibility: Most production networks rely on wired fieldbus and industrial Ethernet. The resulting complex cabling creates challenges for enterprise network deployment, maintenance, and transformation.
• Difficulty in intelligent upgrading: Legacy equipment with high value and long life cycles, coupled with insufficient computing resources and low intelligence levels, hinders the industry’s digital and intelligent transformation.

To effectively empower this transformation, it is necessary to utilize emerging technologies such as 5G-A and AI to build a new type of industrial digital intelligence foundation.

Towards a New Industrial Digital Intelligence Foundation with 5G-A and AI

5G-A and AI, as two of the most representative technological innovations, are mutually reinforcing each other, emerging as dual engines driving the development of industrial digital intelligence (Fig. 1). On one hand, the introduction of AI unlocks the full potential of 5G-A networks by improving 5G-A transmission efficiency and enabling network intelligence, allowing 5G-A to better serve vertical industries. On the other hand, 5G-A, with its proximity to production operations, delivers enhanced mobility, real-time performance, security, and cloud-edge-end collaboration, thereby providing stronger infrastructure support for AI applications.

As 5G-A, AI, cloud computing, and other emerging technologies become increasingly integrated into industries, industrial digital transformation is giving rise to a wide range of new scenarios and applications, while the restructuring of the industrial ecosystem is accelerating. There is an urgent need for upgrading the underlying computing and network infrastructure.

To address this, the new industrial digital intelligence foundation should adopt a flat three-tier architecture consisting of on-site, edge, and center layers, integrating cloud, network, computing, and applications to enable deep convergence of operational technology (OT), information technology (IT), communication technology (CT), and data technology (DT).

It is characterized by ubiquitous interconnection, deterministic capabilities, computing-network-application integration, openness and intelligence, as well as security and controllability.

• Ubiquitous interconnection

Based on various wired and wireless network technologies such as 5G-A, industrial optical networks, industrial Ethernet, and time-sensitive networking (TSN), a ubiquitous interconnected converged network can be constructed. It will meet the diversified networking requirements of different devices and application scenarios, achieving full connectivity among all key production elements, including man, machine, material, method, environment and measurement, and enabling data connectivity, integration, and interoperability across devices, systems, and vendors.

• Deterministic capabilities

As an enhanced evolution of 5G technology, 5G-A provides commercial-grade deterministic capabilities, including 4 ms low latency with 99.99% reliability, millisecond-level jitter, 24/7 high availability, 1 Gbps uplink bandwidth, on-campus data retention, and 5G LAN support. It also supports lightweight, low-cost RedCap terminals, integrated sensing and communication, and Ambient IoT, better meeting the requirements of industrial production control, flexible production, large-scale equipment access, warehousing management, the Internet of Vehicles, and the low-altitude economy.

• Computing-network-application integration

The intelligent digital foundation enables collaborative computing across cloud, edge, and device. The development of platforms like DeepSeek and Manus is driving a shift in the structure of computing power demand from training-centric to inference-centric. The demand for edge computing power closer to applications has exceeded that of centralized computing power, gradually forming a distributed computing power system with centralized training, diversified inference and edge-oriented deployment. Computing power collaboration relies on a cloud-edge-end computing network to perceive, measure, schedule, and control distributed computing resources, achieving unified access, management, and dynamic scheduling to better meet application needs and improve resource utilization.

Based on 5G-A all-in-one equipment, the integrated deployment of computing, network and applications enables dynamic resource sharing, allocation, orchestration and scheduling. This facilitates the deep convergence of IT, OT, CT and DT, while ensuring that SLAs are met. For example, in production control applications involving PLCs, uncertainties in delay and jitter may cause service interruption. Based on TSN clock synchronization, the 5G-A network globally orchestrates the control data, aligning data transmission timing with 5G scheduling cycles, thus ensuring the continuity of production processes.

• Openness and intelligence

Based on open standards, technologies, and interfaces, a full-stack open architecture enables the decoupling of network protocol layers, device software and hardware, applications and networks, models, as well as training and inference, fostering an open industrial ecosystem.

AI deeply empowers 5G-A networks and comprehensively enhances their connectivity, computing, and O&M capabilities. In connectivity, AI optimizes resource allocation and utilization, enables precise service perception and assurance, and enhances end-to-end deterministic capabilities. In computing, AI accurately predicts demands and schedules resources, achieving coordination and optimization across cloud, edge, and end computing resources. In O&M, AI agents enable networks to be self-aware, self-diagnosing, and self-repairing, while also assisting in fault delimitation and localization through interactive intelligent O&M.

• Security and controllability

Through enhanced capabilities in proactive defense, intelligent perception, and collaborative handling, endogenous security ensures stable, reliable, secure, and trustworthy network infrastructure and data.

Data controllability is achieved by building a computing-network system in which data flow, usage, and security risks are all effectively controlled, thus meeting the industry's stringent security requirements for data.

The Industrial Digital Intelligence Foundation Facilitating Innovation in Intelligent Applications

AI is driving innovation and transformation in various industries. The digital intelligence foundation, deeply integrated with the industrial Internet, has been successfully applied to R&D, production manufacturing, business management, and O&M, empowering enterprises to optimize resource allocation, improve quality, reduce costs, and increase efficiency.

• Industrial Internet platform

The industrial Internet platform connects devices vertically and the enterprise horizontally, establishing data links throughout the product lifecycle. It forms an intelligent closed-loop system for optimized decision-making covering R&D, manufacturing, marketing, service, and the supply chain by integrating technologies, consolidating resources, and accumulating mechanisms. The industrial digital intelligence foundation provides data, scenarios, and computing power for industrial Internet platforms, offering comprehensive support from data collection to highly reliable data transmission and edge-cloud computing power assurance.

• Industrial large models and AI agents

DeepSeek greatly reduces computing power requirements and lowers the application threshold for large models, accelerating their application and implementation in industry. Manus drives the transition of AI from a tool to an autonomous decision-making entity, achieving breakthroughs in the industrial application of AI agents. Large models and AI agents have been widely used in intelligent production scheduling, intelligent production resource allocation, production plan monitoring and execution, and process

document generation, greatly improving industrial intelligence. Through the multi-agent collaboration architecture based on the model context protocol (MCP), enterprise-developed agents can be integrated into the system via external plug-ins, replacing traditional customized enterprise application development with AIGC-based approaches, and significantly improving the efficiency of enterprise application development and delivery.

• Embodied robot

In recent years, embodied robots have developed rapidly and shown broad prospects in industries such as manufacturing, home services, and education. They are well-suited for scenarios that require physical interaction, strong environmental adaptability, and tasks that are difficult to complete efficiently by manpower, such as flexible assembly, sorting, and packaging. They effectively solve the automation challenges of the "last meter" in production and logistics. The industrial digital intelligence foundation provides both cloud and edge computing power for embodied robots, ensuring smooth, high-bandwidth, and low-latency communication between the "brain" (cloud), the "cerebellum" (edge), and the motion control system.

The 5G-A+AI-powered industrial digital intelligence foundation serves as a key infrastructure for new industrialization and as a platform for the deep integration between the digital and real economies. Leveraging policies such as intelligent manufacturing, 5G + Industrial Internet, and AI Plus, it is being fully integrated into industrial intelligent transformation and 5G factory development. Through ongoing practice, it continuously improves and iterates on capabilities across networks, computing power, platforms, security, and other components, achieving innovative integration of technologies and applications and driving symbiotic advancement.

5G-A and AI are still developing and evolving rapidly. They will continue to deeply integrate with the industrial Internet, complementing and empowering each other to build a future-oriented industrial digital intelligence foundation that enables new industrialization. With the vision of becoming a leader in network connectivity and intelligent computing, ZTE is committed to driving the digital economy and working hand in hand  with industry partners to usher in a new era of industrial intelligence.