Qualcomm bets on 'Personal AI' as the next growth driver in the terminal market

According to a white paper commissioned by Qualcomm and produced by IDC, global smartphone shipments fell 13.9% year-on-year in the first half of the year, yet the proportion of AI-capable phones reached 66.4%, up 21.8%. PC and tablet shipments also declined, but their AI penetration increased. The white paper argues that the old model of 'stacking parameters for sales' is no longer effective. Instead, it proposes a 'Personal AI' paradigm, where AI agents run on devices centered around the user, sharing a unified understanding across phones, PCs, and glasses. To overcome cost, latency, and privacy bottlenecks of pure cloud AI, Qualcomm advocates a distributed edge-cloud architecture. Tasks are dynamically allocated: lightweight on-device, moderate on edge, complex on cloud. Qualcomm's technical approach includes an on-device personal knowledge graph, a sensor hub for always-on low-power sensing, and multi-core collaboration (CPU, NPU, GPU, sensor hub) to balance performance and power. Qualcomm believes the next-generation entry point will not be a single device but all devices – phones, PCs, glasses, cars – each acting as an endpoint for the same personal AI agent.