Skate to Where the Workload Is Going

By 2027, most organizations expect AI PCs to be the majority of their fleet — and roughly half expect their entire fleet to be AI-capable a year later. Yet the same IT leaders are walking into a memory-driven price spike and the most volatile PC supply environment since the pandemic. The instinct will be to stretch the fleet and wait. The research says that’s backward: the organizations that absorb cost shocks, OS migrations, and security pressure run a steady, rolling refresh — not a heroic one-time buy meant to last forever. This conversation takes the long view on what that means, and why Snapdragon-powered PCs are built for it: how a silicon-level decision about where AI work runs — a dedicated NPU versus loading everything onto the CPU and GPU — plus on-device-vs-cloud economics and platform support commitments compound across every refresh tranche. We’ll forecast what an IT director should specify today for workloads, and AI models, that haven’t shipped yet.