Nvidia Stock Just Did Something for the First Time in 7 Years. Here's What History Says Happens Next.

Nvidia shares are up only ~5% year-to-date in 2026 as the stock’s P/E ratio compresses to the lowest level in seven years, signaling a “digestion” phase after a ~1,100% run since ChatGPT’s 2022 launch. The article flags risks that AI hyperscalers may slow capex and that AMD accelerator architectures plus Broadcom custom ASICs could pressure some data-center workloads. It argues that P/E compression may be normalization rather than business deterioration, citing Nvidia’s expansion beyond GPUs into adjacent AI-stack layers (networking, CPUs, optical interconnects) that could widen the earnings base as Blackwell revenue materializes.

Analysis

The real signal is not that NVDA is “cheap”; it’s that the marginal AI dollar is moving down the stack. When the category leader stops multiple-expanding, capital typically rotates toward picks-and-shovels with higher content per deployed cluster and less dependence on one chipset cycle. That creates a better relative setup for MRVL, COHR, and LITE over the next 1-3 months if hyperscaler capex remains intact, while AMD and AVGO stay exposed to a more crowded narrative around “good enough” alternatives and custom silicon substitution.

The key risk is a capex air pocket, not valuation. If cloud buyers slow orders because inference utilization or enterprise monetization lags, the first de-rating should hit the highest-beta AI supply-chain names, with AMD especially vulnerable because its upside case still leans on share gain rather than proven earnings durability. The falsifier is a fresh round of hyperscaler capex acceleration or stronger Blackwell/adjacent revenue commentary; that would likely turn this from a digestion phase into a re-rating trade within one earnings cycle.

Contrarian view: the market may be over-reading P/E compression as deteriorating fundamentals when it is often just denominator expansion after an extreme move. If earnings continue compounding, the multiple floor can rise again, but only if the company demonstrates monetization beyond the core GPU cycle. Absent that proof, NVDA can stay range-bound even with solid fundamentals, making this more attractive for relative value than outright momentum chasing.

Key Decisions for Investors

• Avoid chasing NVDA into strength; wait for the next earnings/capex readout. If Blackwell demand and hyperscaler spend re-accelerate, use a 6-12 month call spread to express a re-rating view with limited downside.
• Long MRVL or LITE on pullbacks over the next 1-3 months. These names have cleaner leverage to AI networking/optics content growth and less direct competition risk than GPU/CPU peers; upside is strongest if cluster build-outs stay on schedule.
• Pair trade: long COHR/LITE basket vs short AMD for 1-3 months. Thesis: the market is underpricing non-GPU content expansion while still paying for AMD share-gain hopes; thesis fails if AMD wins a materially better design-win cadence or cloud capex slows broadly.
• Set a watch item on hyperscaler capex revisions and AI infra commentary over the next earnings season. Any downward revision is a trigger to reduce AI-beta exposure across NVDA, AMD, and AVGO; any upward revision is a signal to add on weakness.