CPUs Repriced: AI Infrastructure Broadens Beyond GPUs

WisdomTree argues the AI investment cycle is expanding beyond GPUs, with CPUs re-emerging as a critical layer in both training and inference as workloads become more complex. 

• Hyperscaler capex is set to approach $700bn in 2026, sustaining demand across the semiconductor stack.
• CPUs are increasingly essential for data orchestration, reinforcement learning, and agentic AI workflows, where coordination—not just compute—matters.
• Intel and AMD are seeing renewed momentum, though competition from custom ARM chips and Nvidia’s CPU ecosystem is intensifying.

AI was never just a GPU story. The question now is how much of the next leg belongs to everything around it.