More VRAM. Higher costs. Fewer units. Less on-device AI. The memory crisis is not just a pricing problem. It is reshaping where AI workloads run.

Phil Solis, IDC's Research Director for Semiconductors and Enabling Technologies, maps the full AI accelerator spectrum — from cloud to wearable — and what the crunch means for chip strategy. This excerpt gives you the framework.

What You'll Learn

  • How AI workloads span cloud to edge to wearable — and what that means for chip design
  • Why the memory crisis is pushing AI back to the cloud
  • How the NPU market is evolving from integrated to discrete
  • Why discrete NPU volumes remain low today
  • What drives the on-device AI rebound when the crunch ends
  • How privacy, latency, and token costs shape the next wave of AI acceleration

Download the Presentation Excerpt

Available in PDF format

Ready to Go Deeper?

This excerpt gives you the framework. IDC gives you the full accelerator market data, NPU forecasts, and direct analyst access to navigate the next move with confidence.