Intel's blowout quarter showcases the exact reason why we took a stake in Arm

The market is beginning to price a structural re-rating of the CPU layer, not just a one-quarter earnings beat. That matters because CPUs are the control plane for inference-heavy workloads, so the second-order beneficiaries are vendors with broad platform attach, not just the headline AI GPU winner set. If the mix really shifts toward orchestration and agentic workloads, the supply bottleneck moves upstream into general-purpose compute, which should support sustained pricing power for ARM/INTC and improve TAM visibility for AMD over the next 2-6 quarters.

ARM is the cleanest expression of the thesis because it has multiple monetization vectors: royalty leverage from device proliferation plus optionality from its own silicon. The market is likely underestimating how much credibility a successful in-house CPU design gives ARM in negotiations with hyperscalers and OEMs; even modest design wins can change the royalty conversation across the ecosystem. The flip side is that expectations are now moving faster than product-cycle reality, so any delay in qualification, performance, or customer expansion could trigger a sharp reset despite the strong narrative.

The more interesting contrarian angle is that this is not purely a GPU-to-CPU rotation; it is a broader capex reallocation toward infrastructure efficiency. That could pressure GPU-only sentiment at the margin while still leaving NVDA structurally fine, but it also opens a path for incumbents with strong x86 franchises to re-accelerate share if they can prove inference efficiency and platform stability. Over a 3-12 month horizon, the trade is less about absolute AI demand and more about who captures the orchestration budget as inference scales and agentic compute becomes mainstream.