Intel Earnings Blowout Raises Questions Around a 117x Forward P/E

The market is not just re-rating Intel’s earnings power; it is repricing the CPU’s role in AI architecture. If inference and agentic workloads continue shifting compute from concentrated training clusters into distributed orchestration layers, Intel’s socket share can expand even without winning the headline accelerator war. That creates a second-order beneficiary set: hyperscalers and enterprise OEMs that need lower-cost inference infrastructure, while GPU vendors face a slower marginal growth rate in system-level CPU attach.

The real tell in the print is pricing power plus mix shift, not the beat itself. Double-digit ASP inflation in server CPUs tends to be durable only when supply is constrained and OEM/customer validation cycles are already locked, which means the next 2-3 quarters matter more than the one-day spike. If management can hold margin expansion while capex stays heavy, the equity can keep working; if yields or demand soften, the market will quickly punish the stock because the current valuation assumes both operating leverage and flawless execution.

The bear case is no longer “Intel is broken,” but “Intel is finally good enough that expectations are now too high.” At ~117x forward earnings, the stock is pricing a multi-year straight line of improving gross margins, foundry credibility, and sustained CPU share gains. That leaves the name vulnerable to any disappointment in 18A ramp, external foundry conversion, or a reversal in server pricing once inventory normalizes.

Second-order losers include alternative CPU architectures and custom silicon programs that rely on minimizing x86 attachment; the more Intel becomes the default control plane for AI systems, the harder it is for those ecosystems to bypass it. The contrarian setup is that the stock may have overshot near term even if the long-term thesis is valid: the best trade may be a pullback buy on proof of sustained guide-through rather than chasing the gap higher.