Nvidia Stock Has 1 Problem. Here's Why I'm Buying Anyway.

Nvidia posted fiscal 2026 revenue of $215.9 billion, up 65%, with data center revenue of $193.7 billion and sovereign AI revenue topping $30 billion, more than tripling year over year. The article highlights a key risk from hyperscaler customer concentration, but notes that about 40% of revenue now comes from outside the top five hyperscalers and that Nvidia still holds an estimated 80% to 90% share of the AI accelerator market. Overall tone is constructive on fundamentals despite growing competitive and customer-concentration risks.

Analysis

The key second-order issue is not whether Nvidia remains the dominant accelerator vendor today, but whether hyperscalers are converting capex from merchant GPU purchases into vertically integrated silicon over a multi-year horizon. That shift does not need to fully displace Nvidia to matter: even a 10-15% substitution of hyperscaler workloads over the next 12-24 months would pressure mix, pricing power, and the premium multiple on “infinite TAM” assumptions. The market is likely still underestimating how quickly internal chips can cannibalize marginal growth in the highest-volume, lowest-friction accounts.

Counterintuitively, the broadened customer base is more valuable for de-risking than for near-term acceleration. Sovereign and enterprise demand is stickier, less optimized, and less likely to self-build, which should improve revenue durability and reduce headline concentration risk. The trade-off is that these customers typically buy in smaller, more fragmented deployments, so revenue quality improves while growth rate and operating leverage likely normalize over time.

The better near-term setup is for the ecosystem, not just NVDA. If sovereign AI spending continues compounding while hyperscaler custom silicon ramps, demand should diversify into networking, power, advanced packaging, and memory bottlenecks where capacity remains constrained; that can support semi breadth even if NVDA’s share of incremental AI spend eases. The main reversal risk is a capex digestion phase: if the big four slow spend after 2026 buildouts, the entire AI complex could de-rate quickly because forward estimates still assume sustained infrastructure intensity.

Consensus appears too complacent on durability and too pessimistic on substitution timing. Nvidia can remain the default platform for years and still see its growth inflect down meaningfully as customer concentration improves in the wrong direction for the stock. The right framing is not ‘Nvidia is broken,’ but ‘the stock may be pricing scarcity rents that become less scarce as the largest buyers internalize more of the stack.’

Key Decisions for Investors

• Trim NVDA into strength over the next 2-4 weeks; keep a core position but reduce exposure where the stock is trading on peak narrative multiples rather than peak growth certainty. Risk/reward worsens if hyperscaler capex headlines soften, with downside likely to come faster than the fundamentals.
• Pair trade: long GOOGL / short NVDA for 3-6 months as a hedge against custom TPU adoption cannibalizing merchant GPU demand. This expresses the substitution thesis without needing a broad AI selloff.
• Add/selectively overweight AMZN and MSFT on 6-12 month horizons as diversified AI spend beneficiaries, but prefer them only on pullbacks. Their internal chip efforts are a strategic hedge, not a direct accelerator earnings lever.
• Buy call spreads on SOXX or a basket of AI infra names for 6-9 months to express continued AI capex, while limiting single-name risk from NVDA multiple compression. Best if the market keeps rewarding infrastructure spend but rotates away from the single-name winner.
• Avoid chasing NVDA above prior highs without a new catalyst; use any 5-10% drawdown as the preferred entry for tactical longs, because the stock remains structurally strong but is increasingly vulnerable to duration risk in 2027-2028 estimates.