Huawei Outlines Ten-Point Roadmap for an Agent-Centric AI Future

The report argues that the digital landscape is moving from application-centric to agent-centric, where autonomous systems perceive environments and make real-time decisions. With agent traffic expected to dominate over 90% of network activity, traditional cloud and infrastructure models will face unprecedented strain. To bridge this gap, Huawei proposes a roadmap focused on token-centric infrastructure, spanning generation, delivery, and security.

Key pillars of this transformation include scaling computing clusters by 100-fold while reducing task costs by 1,000-fold, and implementing novel memory systems capable of causal accuracy. David Wang, Huawei’s Rotating Chairman, emphasizes that the industry must shift toward SuperPoD systems and agent-native operating environments to ensure these autonomous systems can evolve safely. Parallel to this, the Global Digitalization and Intelligence Index 2026, produced alongside Tsinghua University, projects that AI will generate $27 trillion in cumulative economic value over the next five years, provided nations successfully integrate computing, networking, and storage into their core economic frameworks.