Nvidia Earnings Blow Everyone Away

Nvidia reported 106% year-over-year revenue growth and guided to roughly 70% growth in the coming year, versus Wall Street expectations of about 45%, sending shares up about 7%; management indicated it can meet only around 70% of current demand. The outlook implies top-five hyperscaler capex could rise to $1.3 trillion from roughly $800 billion, supporting AI infrastructure suppliers while higher memory costs pressure consumer-device vendors. CrowdStrike shares rose about 18% after a 26% revenue increase, 51% growth in net-new ARR, a 49% increase in backlog to more than $10 billion, 350bps adjusted operating-margin expansion, and increased full-year guidance, though its 37x-sales valuation and stock-based compensation remain key risks.

Analysis

The investable read-through is a bottleneck rotation rather than a broad semiconductor beta call. Incremental AI spend should accrue disproportionately to HBM/advanced-packaging inspection, high-speed optical interconnect and data-center electrical integration; CRDO, ONTO, TSM and IESC have more operating leverage to unit growth than NVDA once accelerator supply is allocated. Over the next 1-3 months, order commentary from memory vendors, TSM advanced packaging capacity, and hyperscaler capex budgets should determine whether this propagates into estimates or remains a single-vendor demand signal.

The principal 6-18 month risk is that supply additions convert current scarcity into a capex-cycle air pocket. NVDA's gross-margin trajectory is the key early warning: sustained erosion beyond the expected input-cost effect would imply suppliers are gaining share of the economics, while an abrupt recovery in accelerator availability without matching cloud monetization would pressure the entire infrastructure basket. IESC also carries project-timing and labor-execution risk; its apparent power-constrained upside is less durable if utility interconnection delays defer campus construction.

CRWD's setup is more nuanced than the headline growth signal. Flexible consumption contracts can pull module adoption forward and make ARR acceleration look stronger before customer budgets prove durable, while dilution means non-GAAP operating leverage does not fully accrue to shareholders. Local AI is not a cloud-displacement thesis in the near term; it expands endpoint and data-governance complexity, favoring platform security vendors, but raises component-cost risk for device-heavy names such as GRMN before they can pass through price increases.

Consensus is likely underestimating the divergence within AI beneficiaries: the highest-quality second-order suppliers may see upward estimate revisions with less valuation dependence than the leaders, while consumer hardware faces a negative memory-cost shock. Conversely, the market may be overextending the immediate capex extrapolation; any reduction in hyperscaler capex guidance or evidence that AI workloads are shifting toward lower-cost inference would quickly compress the most momentum-sensitive infrastructure multiples.

Key Decisions for Investors

• Initiate a 3-6 month basket long CRDO, ONTO and IESC, sized smaller in IESC for execution risk. Target a 15-25% relative return versus SOXX if hyperscaler capex commitments translate into supplier backlog; exit or reduce if TSM advanced-packaging commentary weakens or hyperscaler aggregate capex guidance is cut.
• Maintain NVDA exposure but do not chase a post-results gap; add only on a 8-12% pullback or after confirmation that next-quarter gross margin stabilizes. The thesis is falsified if gross margin declines materially beyond management's input-cost explanation or if lead-time commentary normalizes before revenue expectations reset.
• Express cybersecurity exposure as a market-neutral long ZS / short CRWD pair over the next two reporting cycles, with equal-dollar sizing. ZS offers similar zero-trust and AI-security demand exposure at less dependence on sustained premium multiple expansion; cover the CRWD short if net retention, cash flow conversion, or module adoption continues to accelerate without a corresponding rise in dilution.
• Avoid adding broad consumer-device exposure for the next 1-2 quarters; consider GRMN as a relative short against a long AI-infrastructure basket only after verifying memory-cost pressure in supplier pricing. Cover if Garmin demonstrates price realization sufficient to preserve gross margin, or if memory spot prices reverse.
• Set a capex-risk alert around upcoming cloud-provider guidance: a sequential slowdown in data-center capex or disclosed lower utilization would warrant taking profits in CRDO/ONTO/IESC first, as these second-order beneficiaries have the greatest estimate sensitivity to a build-out pause.