Nvidia doubled its quarterly revenue to $96 billion and crushes estimates—CEO Jensen Huang says demand is accelerating

Nvidia reported Q2 revenue of $96.2B (+106% YoY), crushing the $92.2B analyst estimate. It guided Q3 revenue to $91B (+/-2%), well above the Street’s $103.9B expectation as described in the article, and data center revenue came in at $89B vs $85.7B estimates (up from $75.2B last quarter). Non-GAAP EPS was $2.22 vs $2.06–$2.09 expected, supporting a bullish read-through on ongoing AI chip demand even as the stock was flat after hours.

Analysis

This is less a fundamental surprise than a validation that the AI build-out is still constrained by supply, not demand. The immediate winners are the ecosystem names that convert every additional accelerator shipped into more rack-level content: networking, power, cooling, and custom interconnect exposure should keep compounding even if the chipmaker’s own upside gets capped by expectations. That argues for a broader AI-infrastructure basket rather than chasing the most crowded single-name expression.

The stock’s muted reaction is the tell: when great numbers stop re-rating the leader, the market is starting to price in saturation of the easy upside. Over the next 1-3 months, the key catalyst is not another beat but whether hyperscaler capex plans and deployment cadence accelerate enough to keep inference utilization tight; if not, the growth rate can decelerate faster than the revenue base suggests. Watch for any sign that customers are shifting spend toward networking/power infrastructure instead of incremental GPU purchases, which would be positive for suppliers but a headwind to the leader’s multiple.

Contrarian take: consensus is assuming AI spend is a straight line, but the more likely path is a rotation within the stack. If software monetization lags, buyers will demand efficiency, which favors custom silicon, networking, and data-center infrastructure over pure accelerator beta. That makes the risk/reward in the leader less attractive after this print, while secondary beneficiaries may still have upside as the market broadens the trade.

Key Decisions for Investors

• Prefer a 1-3 month long ANET / VRT / ETN basket over chasing NVDA here; the trade expresses AI deployment intensity with less valuation risk and better second-order upside if build-outs continue.
• Initiate a 6-12 month long AVGO vs short NVDA pair: AVGO benefits from AI networking and custom silicon diversification, while NVDA’s multiple is more vulnerable if growth normalizes from extraordinary levels.
• Do not add to NVDA on this reaction; wait for a 5%-8% pullback or the next hyperscaler capex read-through before re-entering, because the post-print tape suggests expectations are already very high.
• Set a catalyst watch on the next MSFT/AMZN/GOOGL capex commentary; a pause or moderation there would be the clearest falsifier for the AI-infrastructure momentum trade.
• For more tactical expression, consider a 2-4 month call spread in ANET rather than outright NVDA calls, since the market is likely to reward second-order beneficiaries before it pays up further for the leader.