At the 27th China Hi-Tech Fair held in Shenzhen from November 14-16, 2025, Fan Jianping, Vice President of Lenovo Group and head of its AI Lab, delivered a keynote speech at the China Hi-Tech Forum themed "AI Empowering Future Industries." He predicted that intelligent agents and smart terminals would emerge as the next major trend, with "self-evolving agents" representing a significant milestone in artificial intelligence development.

Fan outlined three key drivers for this evolution: 1) Technological feasibility - Large language models continue improving in both scale and efficiency, with parallel research into compact models. 2) Growing computing power - Tech leaders like Microsoft, AMD and Intel are advancing 40 TOPS computing units to meet market demand. 3) Commercial viability - Addressing customer concerns about privacy and token-based payment models through localized processing.

The Lenovo executive explained that self-evolving capabilities - autonomous adaptation to environmental and task changes - represent a crucial milestone in achieving human-like intelligence. These agents won't just power terminals but will serve as stepping stones toward Artificial General Intelligence (AGI).

During his presentation, Fan traced technological evolution from the PC era through internet, mobile-cloud and cloud-AI phases, noting how each innovation built upon rather than replaced its predecessors. He positioned AI agents as the next transformative wave following the rise of perception AI (2012) and generative AI (2020).

Lenovo has pioneered this transition through: - Developing the world's first AI PC in 2024, now representing 30% of its shipments - Creating a modular agent architecture with six components (brain, tools, knowledge base, etc.) - Implementing hybrid approaches balancing performance with privacy constraints

Unlike academic models requiring massive data updates, Lenovo's solution enables continuous, personalized evolution through: 1) Context-aware adaptation to new scenarios 2) Task-specific optimization 3) Environmental responsiveness 4) Memory-based learning from user interactions

Fan also discussed multi-agent collaboration systems, emphasizing the need for specialized, hierarchical coordination rather than simple parallel processing. He envisioned future human-machine integration through "super agents" capable of complex task decomposition.

As the global PC market leader, Lenovo claims unique capabilities in delivering true edge-AI devices, with its self-evolving AI PC architecture reportedly outperforming competitors' offerings.