• Powers the shift from app-centric to agent-centric AI: As AI evolves from app-based to agent-driven models, 5G Advanced’s (5G-A’s) enhanced uplink capabilities become indispensable to overcome limitations of current 4G/5G networks and enable next-generation services.
• Key technical innovations in 5G-A: Features like intelligent uplink, Single User MIMO, carrier aggregation and smart energy management enhance network throughput, reduce latency and support scalable and energy-efficient AI applications across industries.
• Redefining network evaluation for the AI era: AI-driven services require new evaluation metrics that emphasize QoE to provide more accurate measurement of network performance and new monetization opportunities for operators.
• Transforming devices with cloud-based AI: Devices like China Telecom’s Tianyi 5G-A Cloud Phone leverage 5G-A to offload computing to the cloud, enabling powerful AI features such as large language models and cloud gaming, without requiring high-end hardware.
• Building smart AI cities with 5G-A: Major cities like Beijing, Dubai, Hong Kong and Shanghai are deploying 5G-A infrastructure to power AI applications in transportation, emergency services, smart venues and urban management, achieving impressive speeds and functionality.

AI has been the main focus of telecom industry events for some time, and the recent GSMA Mobile World Congress Shanghai (MWCS) and Huawei’s MBBF were no exception. There is considerable enthusiasm in the industry regarding how AI can drive revitalization, both by optimizing network costs and by increasing revenue through improvements in existing mobile network services and the creation of new ones. This year’s events showcased the latest innovations in 5G network technology and subsequent use cases. The shows were packed with robots, drones, automobiles and other smart devices, giving us a glimpse of what telecom networks will be capable of in the near future.

A key takeaway from this year’s MWCS and MBBF was that AI is shifting mobile interactions from app-centric to agent-centric models, driving a surge in uplink demand. Unlike traditional apps, which rely heavily on downlink speeds, AI agents require high-bandwidth, low-latency uplinks to transmit real-time data for cloud processing. This demands that networks prioritize uplink performance alongside downlink to enable seamless AI services.

Furthermore, the transition to AI-driven networks requires moving beyond traditional speed metrics to adopt a Quality of Experience (QoE) framework that properly quantifies real-world performance for agent-based interactions. However, current networks and evaluation methodologies remain inadequate for these new requirements. This gap makes 5G Advanced (5G-A) the essential solution, as its superior uplink capabilities and other innovations are uniquely positioned to support next-generation AI applications.

Key technical innovations in 5G-A for AI

To meet the demands of AI-driven innovation, 5G-A introduces several critical upgrades:

• Increased uplink throughput: Single User MIMO and Uplink Carrier Aggregation increase peak uplink rates and overall network capacity, enabling massive data transmission for AI processing. Applications such as video streaming will achieve continuous uplink to transmit high-resolution images. Meanwhile, enhanced uplink brings stable and high-capacity connections for smart city deployments, connecting thousands of cameras and IoT devices.
• Better coverage and lower latency: Intelligent Uplink improves uplink coverage capabilities of multiple antennas and combines distributed UCN (User-Centric Network) and time-frequency air control to share all resources in real time, thereby reducing latency, among other benefits. For instance, robots can perform real-time decision-making through instant uplink feedback for obstacle detection and navigation.
• Multi-layer coordination: Optimizing network performance by balancing user loads across different 5G spectrum layers.
• Smart energy management: Stepwise carrier deactivation reduces energy consumption, which in turn increases the battery life.

Redefining network evaluation for AI era: From technical metrics to experience quantification

The emerging AI necessitates a fundamental transformation in network evaluation, moving beyond traditional throughput metrics to a comprehensive AI-QoE framework. A unified industry standard will accelerate mobile AI adoption by securing consistent, high-quality interactions across devices and networks. It will also elevate user experience and unlock new monetization opportunities for operators to provide differentiated services tailored to next-generation applications. The telecommunications industry can effectively evolve networks into active enablers of the AI ecosystem, ensuring technological advancements translate into quantifiable user benefits and sustainable business growth.

5G-A enabling new generation of AI devices

With technical foundations in place, 5G-A is already transforming devices and services. AI smartphones are entering the market in full force and will exceed 70% of all devices sold by 2026, according to Counterpoint Research. 5G-A networks are bringing significant innovations to the device industry, as evidenced by China Telecom’s Tianyi 5G-A Cloud Phone. This device was launched on May 17, 2025, and leverages China Telecom’s 5G-A network built in cooperation with Huawei. The device uses 5G-A capabilities to allow heavy computing to take place in the cloud. Therefore, there is less need for robust computing power on the device itself. This means that a user does not need an expensive device to enjoy a high-quality mobile gaming experience. The device also offers other advanced AI features such as New Calling and China Telecom’s Xingchen large language model.

China Telecom 5G Tianyi 5G-A Smartphone

The Tianyi 5G-A smartphone is powered by 5G-A Intelligent Ultra Pooling Uplink Technology developed by Huawei. This technology employs AI models to predict signal channel quality in real time and also uses AI for time-frequency-RAT-space-power coordinated scheduling, intelligent multi-band selection, RAT decoupling based on service needs, and free uplink scheduling. This technology is not limited to smart devices – it has been deployed in scenarios such as smart transportation, robotics, smart manufacturing and drone management, as it improves 5G network uplink speed by up to 30%.

5G-A enabling AI cities

However, 5G-A isn’t only changing things at the consumer or even the enterprise level. 5G-A is changing the world at the city level as well. Counterpoint Research recently conducted a survey to find the world’s top 100 AI cities, ranked according to a variety of criteria, including the strength of mobile networks. As seen in the chart below, the world’s leading AI cities are deploying the most sophisticated mobile networks, which leverage 5G-A.

World’s Top 5G + AI Cities

Top 15 AI Cities + 5G/AI Initiatives Rankings

It is not a coincidence that the world’s top AI cities are also the pioneers in deploying 5G-A technology. Some examples:

• China Unicom Beijing was the world’s first telco to deploy a large-scale 5G-A network in the country’s capital in late 2024. This network, covering 10 million people, utilized old phone booths to deploy new 5G-A infrastructure to enable services such as ride sharing, emergency services and cloud gaming. At Workers’ Stadium, the 5G-A network enables users to enjoy download speeds of up to 10Gbps and uplink speeds of 4Gbps when the stadium is at its peak capacity of 68,000 people.
• In Dubai, du deployed the largest indoor 5G-A network in the Middle East at the end of 2024. This network uses Huawei’s Three Carrier Aggregation (3CC) technology and relies on the LampSite X digital indoor solution. The network boasts downlink speeds of 5.1 Gbps and enables a variety of indoor solutions such as precise positioning. It will be used for a variety of AI city applications going forward.
• In Hong Kong, Hutchison has partnered with Huawei to deploy 5G-A technology to tackle a variety of issues. The company is deploying the technology in venues such as the HK Exhibition Center, West Kowloon Cultural District and the new HK$3.9-billion Kai Tak Sports Park to provide a superior user experience in crowded venues. The company also recently used 5G-A modems to provide free Wi-Fi at the Art Basel visual arts fair. This network offered speeds of 3.1 Gbit/s downlink and 370 Mbit/s uplink, respectively.
• In Shanghai, the Shentong Metro Group recently announced that it has upgraded 80% of its lines to 5G-A. The rollout spans all 21 metro lines in the city, including the Maglev and airport rail link, and covers 896 km and 517 stations. The network can deliver both downlink and uplink speeds of 600 Mbps, even in tunnels, and peak speeds of over 1 Gbps. The company has gone one step further and has even deployed 5G-enabled inspection robots to monitor track conditions. These robots have replaced over 80% of manual inspections, achieving an accuracy rate of around 98%.

Outlook

AI-driven mobile network innovation has been evolving significantly over the last few years, but more recently 5G-A – particularly its uplink enhancements – has emerged as a key technology which mobile operators can use to unlock a variety of new services, from the consumer level all the way to the municipal level. Operators aiming to deliver superior AI experiences must prioritize 5G-A deployments to stay ahead in the evolving telecom landscape. By integrating advanced network technologies, smarter energy management, and AI-QoE standards, the industry can ensure that 5G networks not only support but actively drive the AI revolution.