Nvidia Roared. Why Didn't All AI Stocks?

Nvidia reported blowout results with quarterly revenue of $96.2B (+18% QoQ, +106% YoY) vs. about $92B expected and EPS of $2.22 vs. $2.08 expected. Guidance was equally strong: $108B revenue (+/-2%) for the current quarter vs. $104B modeled, and CFO Colette Kress guided fiscal 2028 revenue to ~70% growth (vs. ~44% consensus), framed as supply-constrained—implying roughly $670B revenue next year. Shares were up >8% immediately, though the AI complex faces tariff risk as the Trump administration weighs broader semiconductor tariffs that could hinder an AI buildout during a supply-chain reshoring push.

Analysis

NVDA’s print likely reinforces a near-term winner-take-more regime in AI capex, but the second-order effect is margin pressure migrating down the stack. When the leader can raise demand guidance under supply constraints, the scarce inputs — advanced packaging, HBM, substrates, and server integration — become the bottleneck, which tends to compress economics for second-tier hardware assemblers and delay margin recovery for customers that are still capex-heavy. The immediate reaction can stay positive for weeks, but the next leg is more about who captures pricing power than about total demand growth.

The tariff overhang is a different risk path: it doesn’t need to kill AI demand to matter. If applied broadly to servers/laptops/peripherals, it raises landed costs for hyperscalers and enterprise buyers, which can slow deployment cadence by 1-2 quarters even if budgets remain intact; that is bearish for hardware gross margins first and only later for cloud revenue growth. AAPL and GOOGL are not the same trade here — AAPL is more exposed to hardware import friction, while GOOGL is more exposed to capex inflation through data-center build costs and power infrastructure timing.

SVV is the cleaner “AI applier” angle because the operating leverage comes from lower sorting/pricing labor, not from AI sentiment multiple expansion. The thesis is that even modest pricing accuracy gains can lift gross margin over the next 2-4 quarters, but the market likely underestimates execution risk: thrift traffic, wage inflation, and merchandise mix can swamp software gains if consumer demand softens. The contrarian view is that NVDA may be right on demand, but the stock’s upside is increasingly dependent on continued supply tightness; if memory inflation eases or tariffs are diluted, the winner’s scarcity premium can fade quickly.

Key Decisions for Investors

• Stay long NVDA on dips, but treat the trade as a 1-3 month momentum/capex confirmation trade rather than a clean multiple rerate; falsify if gross margin guides below the low-71% area or if next-quarter revenue guidance stops accelerating.
• Pair long SVV / short a basket of hardware-intensive AI beneficiaries over 2-4 quarters: the thesis is lower AI costs widen margins for appliers while tariff and component inflation squeeze the physical-build layer.
• Buy downside protection on AAPL into any tariff headlines; use a 1-3 month put spread as a hedge against broader import-cost pass-through, with the thesis invalidated if policy language excludes finished electronics.
• Watch GOOGL and other hyperscaler proxies as a slower-burn short on capex inflation: if server tariffs broaden, expect deferred deployment rather than lost demand; reduce the short if cloud capex commentary stays intact in the next earnings cycle.
• Set an alert on NVDA implied volatility post-earnings crush: if the stock holds gains while vol normalizes, the cleaner follow-on trade is a relative long in AI appliers like SVV versus the high-beta hardware complex.