Nvidia (NVDA) Q2 2027 Earnings Call Transcript

NVIDIA reported Q2 FY2027 revenue of $96.2B (+106% YoY) and Data Center revenue of $89.0B (+117% YoY), with non-GAAP diluted EPS of $2.22 (+120% YoY) and record shareholder returns of $26.0B ($20.0B repurchases, $6.0B dividends). Management guided Q3 total revenue to $108.0B ±2% and expects FY2028 growth of ~70% YoY, explicitly citing a supply-constrained outlook (including memory price spikes) that will drive gross margins down to a 71–72% range in Q4 before settling at 72–73% in FY2028. Despite the constraint, the company emphasized a platform shift (Vera Rubin plus CPUs and networking) and highlighted AWS deployment of an additional 2M GPUs through Q2 FY2029.

Analysis

The key market read-through is that NVIDIA is no longer just a semiconductor beneficiary; it is becoming the toll collector on AI capital formation. That shifts upside away from pure chip beta and toward infrastructure enablers with balance-sheet capacity and financing expertise: AMZN gains from deep AWS integration, while BLK/BX/KKR/GS/MS monetize the new private-credit/structured-finance layer that is making AI campuses fundable. The underappreciated loser is the middle of the stack — server integrators and generic OEMs like DELL — because more of the economic rent is migrating to the platform owner, while memory and networking suppliers are being squeezed by NVIDIA’s ability to reprice the system around its own roadmap.

Near term, the tradeable risk is not demand but margin sequencing. Gross margin pressure from memory and extended customer payment terms can create a 1-2 quarter earnings-quality headwind even if revenue remains scarce and coveted. The bigger catalyst path over 1-3 months is whether financing announcements convert into funded projects; if they do, the stock stays in a “scarcity premium” regime. If credit markets tighten, or if customers delay land/power builds, the market will start discounting backlog as deferred revenue rather than durable demand. Over 6-18 months, the real question is whether custom silicon programs at large labs and clouds actually reduce NVIDIA content per watt; if not, the platform multiple can stay elevated.

Consensus is likely over-fixating on the top-line beat and underweighting the fact that NVIDIA is now partially substituting its own financing for the market’s willingness to underwrite AI capex. That is bullish while capital is abundant, but it creates a hidden dependency on credit spreads and private-market risk appetite. The setup looks most attractive on pullbacks if management keeps converting financing partnerships into shipped capacity; it becomes fragile if the next read on memory inflation, DSO, or capex deferrals suggests the cycle is being stretched rather than accelerated.

Key Decisions for Investors

• Long NVDA on weakness over the next 2-4 weeks; thesis is that supply scarcity plus financing optionality supports multiple persistence, but reduce if Q3/Q4 gross margin guidance is revised lower again or if DSO keeps stretching beyond 60 days.
• Pair trade: long NVDA / short DELL for 1-3 months. Rationale: the platform owner is capturing more of the AI stack economics while OEMs remain lower-margin pass-through vehicles with weaker pricing power.
• Overweight AMZN versus other hyperscaler proxies over 3-6 months. AWS is a direct beneficiary of the installed-base expansion and should see better AI monetization per dollar of capex than peers that are still funding buildout before harvesting returns.
• Basket long BLK + BX + KKR + GS + MS for 6-12 months as AI infrastructure finance intermediaries. This is a second-order beneficiary basket if structured finance replaces some of the equity burden of data-center expansion.
• Watchlist alert: if credit spreads widen or private funding for NeoClouds stalls, trim NVDA into strength and rotate toward cash-rich hyperscaler/financier beneficiaries. That would falsify the 'scarcity turns into durable growth' thesis.