Nvidia Could Have a Monster October. Here's Why.

Nvidia is projected to deliver 70% revenue growth in fiscal 2028, supported by anticipated AI-hyperscaler data-center capital expenditure increasing from nearly $800 billion in 2026 to $1.3 trillion in 2027. The article argues that upcoming hyperscaler guidance during October earnings could validate Nvidia's demand outlook and prompt a rerating. Nvidia shares, up about 21% year-to-date, trade at roughly 28x trailing earnings and an estimated 14x fiscal-2028 earnings, versus the author's 30x long-term fair-value multiple.

Analysis

The investable event is not a generic AI narrative but the degree to which hyperscaler capex guidance converts into binding 12-18 month accelerator orders. NVDA's upside requires not merely higher aggregate capex, but a sustained mix toward accelerated compute rather than land, power, networking, and custom silicon; that distinction will be visible in commentary from MSFT, META, AMZN, GOOGL, and ORCL. A synchronized capex raise would also tighten the supply chain for HBM and advanced packaging, favoring SK Hynix, MU and TSMC alongside NVDA, while raising the risk that networking and power-infrastructure names have already discounted the spending cycle more fully.

The key contrarian point is that management capex announcements are poor proxies for near-term NVDA revenue unless they include deployment timing and GPU-specific language. Large projects can be delayed by power interconnection, data-center construction, or model monetization constraints, shifting spend into 2028 without changing headline budgets; this would pressure NVDA's earnings multiple even if long-term demand remains intact. Conversely, a capex increase paired with evidence that inference demand is broadening beyond frontier-model training would support a higher terminal earnings base rather than simply pulling forward orders.

Over the next days, guidance language can drive a reflexive rerating, but the more durable 1-3 month catalyst is whether customer comments corroborate order visibility and whether NVDA's next guide implies limited execution friction. For the 6-18 month view, watch gross-margin trajectory: rising supply availability and competitive ASIC adoption could leave revenue robust while limiting incremental EPS and valuation expansion. The thesis is falsified by two or more major customers holding capex flat or shifting spend materially to internally designed silicon, or by NVDA guiding datacenter growth below the rate implied by current buy-side expectations.

Key Decisions for Investors

• Initiate a measured long NVDA ahead of the October hyperscaler reporting window, sized for event volatility rather than a full fundamental position; add only if at least two of MSFT/META/AMZN/GOOGL explicitly raise 2027 capex and link it to accelerated compute. Target a 10-15% rerating over 1-3 months; exit on customer-guidance evidence of delayed deployments or a meaningful deceleration in NVDA datacenter guide.
• Express the supply-chain confirmation trade via long MU or SK Hynix versus short SOXX after capex validation, rather than adding beta to NVDA alone. HBM content and qualification constraints provide a clearer incremental-volume mechanism, while SOXX hedges broad semiconductor risk; reassess if memory pricing weakens or customers signal an ASIC-led mix shift.
• Avoid treating announced capex totals as sufficient confirmation. Set an alert for customer disclosures on GPU procurement, inference capacity, power availability, and depreciation-life changes; absent those details, regard any NVDA rally as sentiment-driven and reduce exposure into its results.
• For a defined-risk event structure, consider a 2-3 month NVDA call spread rather than outright short-dated calls, financed only opportunistically with an out-of-the-money put spread. The likely outcome is elevated implied volatility through clustered customer reports, and a capped structure protects against the material risk of capex being reaffirmed but deployment being deferred.