Nvidia used to benefit from AI spending increases. So why did the stock just slide?

Nvidia shares fell 4.6% on Thursday despite signs that Big Tech is increasing AI-chip spending. Investors appear focused instead on rising competition in the AI processor market, which is weighing on sentiment. The article points to a tension between stronger end-demand and concerns over market-share pressure.

Analysis

The market is pricing a shift from a pure demand story to a margin-and-share story. For NVDA, the issue is not whether AI capex keeps rising, but whether incremental dollars accrue to Nvidia at the same gross margin and share of wallet once hyperscalers push harder on multi-sourcing, in-house silicon, and procurement leverage. That matters because even modest share leakage can overwhelm the optics of accelerating end-demand when the stock is already valued on sustained near-monopoly economics.

The second-order winner is the broader AI supply chain that can capture budget without being tied to a single architecture. Foundry capacity, advanced packaging, memory bandwidth, networking, and power infrastructure should remain better insulated than the headline GPU name if cloud buyers diversify suppliers to improve negotiating leverage. Over the next 3-12 months, competition pressure is more likely to show up first in commentary, lead times, and mix than in absolute unit demand.

The key risk is that investors are underestimating how fast customer concentration can turn into customer bargaining power once capex is large enough. If one or two hyperscalers signal credible internal accelerants, the multiple can compress well before revenue trends break, because the market discounts a future where Nvidia still grows but no longer controls pricing. The contrarian read is that the move may be partially overdone near term: switching costs, software lock-in, and ecosystem inertia make displacement slow, so the better trade may be against further multiple compression rather than against the franchise itself.

Key Decisions for Investors

• Trim NVDA into strength over the next 1-2 weeks rather than chase the dip: keep core exposure, but reduce marginal upside risk while competition headlines are actively repricing the multiple.
• Pair trade: long MRVL or AVGO vs short NVDA for 1-3 months to express continued AI spend with lower single-name competition risk; thesis is budget diversification, not a capex slowdown.
• For event-driven positioning, buy NVDA downside protection 1-2 months out at strikes ~8-12% below spot to hedge a further de-rating if hyperscaler commentary turns to in-house silicon or multi-vendor sourcing.
• Watch for confirmation in the next earnings cycle: if cloud capex stays high but NVDA guidance/GM assumptions stop expanding, fade rallies; if margins and order commentary hold, cover shorts quickly because the stock can re-rate fast on ecosystem stickiness.
• Longer-dated relative-value idea: overweight AI infrastructure beneficiaries such as VRT or ANET vs NVDA over 6-12 months, as the spend shifts from compute concentration toward power, cooling, and networking intensity.