PRESS RELEASE: At the Wireless Media and Analyst Roundtable held at MWC Barcelona 2025, Eric Zhao, Vice President and Chief Marketing Officer of Huawei Wireless Solution, made a keynote speech titled “Intent-Driven AI-Centric 5.5G Towards Advanced Autonomy”, sharing Huawei’s recent achievements and prospects for mobile AI.
“Agentic workflows make intent-driven AI-centric 5.5G network a reality, enabling advanced autonomy to reshape network O&M, user experience, and business models,” said Zhao.
2024 witnessed explosive growth in AI and rapid adoption of AI in mobile communications. In the year, Huawei launched its telecom foundation model, digital twins, and highly autonomous network architecture (HANA) to enable high-level autonomous networks. Its RAN intelligent agent has been deployed in 9 cities, with a total of more than 32,000 sites, and has empowered more than 1,200 wireless network engineers with AI Copilot to help them achieve efficient O&M, performance optimization, and user experience improvement. These achievements set a benchmark for moving wireless networks from the mobile Internet era to the mobile AI era.
Mobile AI Differentiates Connections, Increasing Network Complexity
The rising popularity of AI fosters collaborative innovation between humans and machines, leading to a diversification of services and user experiences. This evolution will significantly increase network complexity. Building intelligent networks to reduce network OPEX has become a core challenge for operators. At the same time, mobile AI enables operators to transition from traffic concentration to multi-factor revenue generation, presenting vast opportunities for business monetization. Efficient network scheduling in response to service needs will be an industry-wide challenge.
Agentic Workflows Are Agent-Native to Enable Advanced Autonomy
Huawei has proposed agentic workflows to help operators fulfill these goals. Agentic workflows move away from software-defined practices to automatic task process generation based on intent-driven translation. Intent-driven networks are agent-native, which optimizes services across the board through multi-agent parallel processing and collaboration. The digital twin systems evaluate service results in real time and report the changes to agents for constant iterative optimization.
Three-Layer AI-Centric 5.5G Network Meets Diverse Requirements
Huawei’s AI-centric 5.5G has a three-layer architecture: RAN intelligent service engine (RISE), MBB automation engine (MAE), and digital sites. With the three-layer architecture, networks translate service demands into intent and make decisions for resource scheduling. Multiple agentic workflows, including GigaGear, GreenPulse, and GainLeap, are already supported.
- GigaGear translates service level agreement (SLA) intent into network requirements and schedules network resources to deliver diverse user experiences.
- GreenPulse focuses on multi-agent collaborative decision-making for optimization across the board, enabling “0 Bit 0 Watt 0 Loss”.
- GainLeap precisely identifies new service requirements and automatically evaluates and deploys resources, which ensures business monetization specific to users, services, and scenarios.
Huawei will continue to develop new agents and workflows based on service scenarios and intent to meet the diverse service needs of the mobile AI era.
Ecosystem Openness Creates a Shared Future for Mobile AI
Huawei will build an iNet community to help expose network capabilities to third parties to facilitate application development. The community provides powerful model capabilities, massive special tokens, and diverse APIs, helping define more scenario-based agentic workflows in collaboration with partners to create more business opportunities.
AI will help operators generate more value out of their networks. Huawei will continue to work with all operators and industry partners to seize opportunities and achieve business success in the mobile AI era. Zhao concluded, “We believe AI-centric 5.5G, fueled by innovation and ecosystem openness, will create infinite possibilities in a new chapter of advanced network autonomy.”