Huawei Shapes the AI-centric Network Future

Rather than being just a “hot topic” for the next few years, AI will impact every aspect of every user’s life and be adopted by mobile operators as a productivity tool to create value, reduce costs and boost network ROI, according to Chief Marketing Officer of Huawei Wireless Networks, Zhao Dong.

Speaking at Huawei’s Wireless Summit at MWC2025 in Barcelona, Zhao set out Huawei’s vision for a flexible, intent-driven intelligent network built on the convergence of 5.5G (5G-Advanced) and AI technologies, and addressed the opportunities and challenges associated with creating the AI-centric network.  

Zhao pointed to three major challenges created by the convergence of networks and AI - how to make networks more flexible in order to meet the complex requirements associated with AI applications, how to manage the higher opex demands associated with the introduction of network AI; and how to open up the opportunities created by AI and allow operators to monetize new services. “We all need answers to these three challenges,” he said.

Autonomous Networks and the Intelligent RAN

Zhao said that as wireless network intelligence evolves the industry is moving closer to a highly autonomous network architecture (HANA) through initiatives such as the Telecom Foundation Model (TFM) and RAN digital twinning, which are reshaping operators’ experience in terms of network O&M, service creation and customer experience, in the AI world. The company’s intelligent RAN solutions, already widely deployed across nine cities in China, had encouraged Huawei to accelerate the process, he added.

As mobile AI sets higher expectations for RAN intelligence the industry needs to discuss how to support AI applications across all scenarios, said Zhao. The advent of the AI-centric network model, specifically with regard to 5.5G, is enabling this greater degree of autonomy. Revenues that have traditionally been derived from a single service proposition will be opened up with the incorporation of AI, to a far more diverse range of service experiences.

Today’s software-defined networks (SDNs) are not truly autonomous as they only addresses Levels 2 and 3 of the network, said Zhao. “Our vision is to upgrade all features in the network to achieve Level 4 or Level 5 autonomous networking through intent-driven capabilities, with the defined rules of today’s SDN architecture being replaced by an intelligent and highly adaptive model, so contributing to opex reduction.” Introducing agentic workflow in a closed-loop structure across the planning, analysis, decision-making and execution phases will represent a totally new autonomous network model, said Zhao.

Threefold opportunities of AI adoption

In order to reap the benefits of AI and bring higher value to their pipes, telecom operators needed to plan their networks to address the diverse needs of AI services, said Zhao. The major benefits of this approach will be increased spectral efficiency, greener and more energy efficient networks, enhanced O&M capability, and the capacity to address the diverse demands of AI services.

When Huawei designs its new 5.5G network solutions it includes provision for AI adoption and the needs of AI services, Zhao explained. AI will bring the benefit of simplifying O&M in a more complex 5.5G environment. ”We’d like to adopt AI to increase the choreography and the collaboration of AI services, so that network resources will feature higher efficiency and higher flexibility. In the future we look forward to working together with operators to seize this opportunity. Huawei is willing to welcome industry partners to enable more business opportunities for agentic AI.”

Zhao went on to highlight three key areas of opportunity he believed would flow from AI-centric adoption - traffic simulation, device form factor and person-to-network interaction, and the synergy between devices and the cloud.

On traffic simulation, Zhao noted that the generation of tokens on networks as a result of GenAI is already consuming more and more data in the pipe, and will account for an increasingly large proportion of mobile data in the future. Last year in China alone the amount of traffic on the network generated by tokens grew by 33 times, while the traffic generated by paid tokens grew by 15 times, said Zhao. However, cost savings in excess of 90 per cent as recently demonstrated by DeepSeek’s AI model would be good news for future mobile networks.

Secondly, AI will change device form factors and the way that people interact with networks, said Zhao. New applications such as personal co-pilots running on digital devices can increase efficiency as well as mobile traffic. AI will also change the way people interact with networks, allowing them to not just consume data from networks, but for the interaction between people and the network to evolve to multimodality. Interaction with networks will be through voice and video, while AI will try to understand the semantic difference between data as well.  This multimodal interaction between people and networks will raise the demanding requirements on mobile traffic and the bandwidth.

The third opportunity lies in the synergy between devices and cloud, said Zhao. Terminal devices are limited by the battery life and the screen size so that it’s impossible to complete all the inference (machine learning tasks) on devices. Therefore some of the inference tasks will have to be offloaded to the cloud, and that means that interaction between device and cloud will increase.

The three opportunities are also challenges for telcos because telcos are supposed to re-shape their networks to offer better coverage, uplink and latency, only thus can telcos serve the new needs in the AI era. The 5.5G solutions designed by Huawei are addressing these changes for telcos, so that they’re in a better position to embrace the AI era, Zhao explained.

Working to remove pain points

Also speaking at the event, Calvin Zhao, President of Huawei’s MAE Product Line, referenced the combination of 5.5G and AI already being used in specific application areas such as healthcare, as well as to power efficiency upgrades in applications such as mines and ports. He explained that allowing vendors to quickly integrate their capabilities into the telecom operators’ processes is improving overall efficiency and energy consumption for telcos, and in this way AI and 5.5G are encouraging the development of each other.

Huawei is understood to be working with China Mobile to develop and mature these features, so letting the value manifest sooner. In the future, Huawei says it will continue to work with  operators in China, the Middle East and Europe.

Zhao Dong said Huawei’s aim is to address the challenge of increasing costs and limited revenues that have been common pain points for the industry. In the past, new radio access technologies or new spectrum bands have required new investment, whereas AI and 5.5G are encouraging the development of each other. “If telcos want to achieve revenue growth they must embrace new opportunities in the market….and to embrace these opportunities by upgrading networks,” he said.

.