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

Nvidia reported fiscal 2026 revenue of $215.9 billion, up 65%, with data center revenue of $193.7 billion and sovereign AI revenue above $30 billion, more than tripling year over year. The article highlights a key risk: over 50% of revenue came from the top five cloud providers/hyperscalers, while two direct customers accounted for 36% of fiscal 2026 revenue. Overall, the piece is constructive on Nvidia's growth and moat, but cautious about customer concentration and hyperscaler in-house chip development.

Analysis

The key second-order issue is not whether Nvidia remains the dominant AI accelerator supplier, but whether its growth rate increasingly depends on a narrower set of buyers while the rest of the market becomes more price-sensitive. Hyperscaler custom silicon is a gradual substitution, not an overnight displacement, which means the bigger near-term risk is margin compression from customer bargaining power rather than a sudden unit collapse. That dynamic is often underestimated because the market treats any “share gain” in AI capex as structurally sticky, when in reality the buyers with the largest wallets are also the ones with the best incentive to internalize the stack.

The offset is that the addressable demand pool is widening into sovereign and enterprise deployments that are less capable of designing their own alternatives and more likely to pay for a turnkey platform. That mix shift matters because it can reduce concentration risk even if it also lowers average customer sophistication and may pressure ASPs over time. In other words, the bull case is less about hyperscalers staying loyal and more about Nvidia becoming the default procurement layer for everyone else who cannot build in-house.

The consensus is probably still too focused on unit growth and not focused enough on durability of pricing power over the next 12–24 months. The market can support a very large multiple while growth is above 40%, but once hyperscaler capex normalizes or internal chips hit production scale, the question becomes whether Nvidia is a monopoly-like platform or just the highest-end merchant supplier. That transition risk is why the stock can stay strong even as the risk/reward gets less asymmetric from here.

A practical catalyst map: the next 1–2 quarters are about order visibility and whether non-hyperscaler revenue continues to offset any moderation in cloud demand; the 6–18 month window is about TPU/ASIC adoption and whether those programs stay internal or spill into broader workloads. If custom silicon begins displacing inference first, the market may not see it immediately in headline revenue but will in gross margin trajectory and inventory behavior. That is the tell to watch.

Key Decisions for Investors

• Stay long NVDA, but reduce conviction sizing versus prior levels; use any post-earnings strength to trim 10-15% into strength if implied growth begins to slow, because the stock is now more exposed to multiple compression than outright fundamental disappointment.
• Pair trade: long NVDA / short INTC only tactically if you want relative AI exposure; the better version is to overlay a hedge with short SOXX calls or a small short basket of hyperscalers on signs of capex moderation, since the second-order risk is spending normalization across the whole complex.
• Buy NVDA downside protection 3-6 months out via put spreads into any rally, targeting a hedge against a 10-15% drawdown if hyperscaler commentary shifts from expansion to optimization; risk/reward is attractive because volatility should remain elevated around capex guidance.
• Add exposure to sovereign/enterprise beneficiaries rather than chasing the core AI winners: prefer NVDA on pullbacks, but pair it with names leveraged to non-hyperscaler AI deployment where custom silicon is less of a threat, as a way to preserve upside while diversifying the concentration risk.
• Monitor GOOGL, AMZN, META, and MSFT capex and ASIC disclosures as the leading indicator; if any of the four show accelerating internal chip utilization, rotate out of high-beta semiconductor beta and into a more defensive software/infrastructure mix over the following 1-2 quarters.