3 Reasons to Buy Nvidia in the Fall

Nvidia guided for $108 billion of current-quarter revenue, implying 89% year-over-year growth, after revenue growth accelerated to 106% in fiscal Q2 2027. Wall Street EPS estimates have risen to $9.31 for fiscal 2027 and $15.68 for the following year, increases of 43% and 85% from estimates at the start of the year; the shares trade at roughly 14x next-year earnings. The article views a potential easing of U.S.-China restrictions on data-center chip sales as upside, while citing unsustainable 75% gross margins, AI-safety concerns, and data-center opposition as key risks.

Analysis

The relevant question is not whether NVDA screens cheaply on a forward estimate, but whether consensus has correctly priced the durability of the hyperscaler capex cycle. A low-teens multiple only holds if forward EPS is credible; it becomes optically misleading if customers shift from frontier-training clusters to lower-cost inference architectures, utilization disappoints, or Blackwell/Rubin supply normalizes faster than demand. The next earnings print is therefore less about a headline beat and more about backlog conversion, customer concentration, networking attach rates, and gross-margin trajectory.

Export-control optionality should be assigned little near-term value. Any détente that permits materially compliant shipments could be positive for NVDA, but it would also validate domestic substitution efforts and invite tighter political scrutiny; the more durable beneficiary of restricted access may be Chinese accelerator and networking supply chains rather than NVDA. Conversely, a further tightening of controls would affect not only lost revenue but also the market's assumed terminal growth rate, creating multiple compression disproportionate to the direct sales exposure.

Second-order exposure favors memory and power/thermal infrastructure if AI cluster deployments remain broad: HBM suppliers and data-center electrical equipment can retain pricing power even if accelerator ASPs eventually compress. AMD is the cleaner competitive hedge: its upside requires only incremental share gains or a credible inference alternative, whereas NVDA must sustain both dominant share and exceptional economics. The contrarian view is that the market may be underestimating margin normalization rather than revenue growth deceleration; a modest gross-margin reset can matter more to EPS than another quarter of revenue upside.

Near term, the stock can remain supported into results if enterprise demand evidence broadens beyond the largest cloud buyers. Over 1-3 months, watch for hyperscaler capex guidance and evidence that deployment bottlenecks are moving from GPUs toward power, grid interconnects, and data-center permitting. Over 6-18 months, the key risk is that physical infrastructure constraints defer revenue recognition and turn apparent demand into extended backlog rather than incremental earnings.

Key Decisions for Investors

• Maintain a core long NVDA only into earnings with a defined trim discipline: add on a post-results pullback if revenue guidance rises while gross-margin guidance remains stable or improves; reduce if gross margin falls materially without a compensating acceleration in networking/software mix. Treat a guide-down in hyperscaler capex as the primary thesis falsifier.
• Express relative AI demand strength via long NVDA / short AMD at roughly beta-neutral sizing over the next 1-3 months, but reverse or exit if AMD demonstrates sustained accelerator share gains and improving MI-series software adoption. This isolates execution leadership from broad AI multiple risk.
• For a less crowded implementation, monitor HBM and data-center power/thermal beneficiaries rather than adding to the accelerator trade after a sharp NVDA rally. Initiate only after confirming order-book growth and capacity discipline; these names benefit from cluster scaling but face cyclicality if deployments slip.
• Do not underwrite China revenue in base-case estimates. Set an event alert around formal export-policy changes; only add an NVDA tactical upside position after explicit rule changes and customer/order visibility, since diplomatic optics alone have low conversion value.
• Use NVDA put spreads rather than outright shorts to hedge a portfolio-wide AI exposure through the next earnings cycle. The asymmetric risk is a margin or capex-duration disappointment causing multiple compression despite a revenue beat; exit the hedge if capex guidance broadens across cloud customers and margin resilience is confirmed.