Why I Keep Buying NVIDIA Stock—And Won't Stop Soon

NVIDIA posted Q4 FY2026 revenue of $68.13B, up 73.2% year over year, with non-GAAP EPS of $1.62 versus $1.52 consensus and net income up 94.47% to $42.96B. FY2026 revenue reached $215.94B, free cash flow was $96.58B, and the company returned $41.1B to shareholders while still authorizing $58.5B in buybacks. Management guided Q1 FY2027 revenue to $78.0B, though China-related compute revenue was excluded and the stock faces export-control risk.

Analysis

The market is still treating this as a single-name AI winner, but the second-order trade is broader: NVDA’s backlog converts directly into leverage for TSM, rack integrators, networking, and power/cooling infrastructure, while compressing the optionality of every “AI platform” claim that lacks a hardware entitlement. If the spend curve holds, the next leg is less about model launches and more about industrial bottlenecks—substation capacity, liquid cooling, packaging, and HBM supply—so the winners extend beyond semis into the picks-and-shovels layer around data centers.

The biggest near-term risk is not demand, it is sequencing. A China write-off can coexist with record growth, but it also lowers the bar for future quarters and makes the stock more sensitive to any hint of a digestion phase once hyperscaler capex lapses even modestly. That matters over the next 1-3 quarters: if customers are pre-paying for capacity now, the market can misread normalization as “demand slowdown” and de-rate the multiple before the next product ramp re-accelerates revenue.

Contrarianly, the consensus may be underestimating margin durability but overestimating narrative duration. The right question is whether Rubin meaningfully widens the performance-per-watt gap fast enough to justify another capex wave, or whether customers simply re-time orders to wait for better economics. If token-cost reduction is real, it expands TAM and supports a longer cycle; if it is mostly a price/performance reset, suppliers may face a temporary air pocket as buyers pause to re-optimize architectures.

Key Decisions for Investors

• Stay long NVDA into the next 1-2 quarters, but finance part of the position with covered calls 10-15% OTM to harvest elevated sentiment premium while protecting against a post-earnings digestion move.
• Pair trade: long TSM / short a basket of software names with weak AI monetization claims over 3-6 months; the hardware supply chain has clearer entitlement to near-term capex than application-layer stories.
• Add META on pullbacks only if capex guidance remains upward-revised; the trade is a latency bet on monetizing AI infrastructure, not a pure multiple expansion story.
• Consider a relative long CRWV vs short a broader cloud/hosting peer basket for 6-12 months if financing conditions remain loose; the market is likely to keep rewarding exposed capacity providers as the buildout deepens.
• Avoid chasing SNPS outright; use it as a hedge candidate against a semicap rerating because design-tool demand can lag if customers delay next-gen tape-outs while awaiting post-Rubin economics.