Nvidia’s Huang expects to sell twice as many chips next year

Nvidia CEO Jensen Huang said the company could sell twice as many chips over the coming year if supply were sufficient, underscoring exceptionally strong AI-chip demand. Nvidia's August guidance projected 70% revenue growth; the potential doubling refers to achievable chip sales under improved supply availability rather than a formal revised revenue forecast.

Analysis

The investable signal is not the aspirational unit-demand statement; it is whether Nvidia converts demand into shippable systems without a mix or gross-margin concession. The binding constraints are likely advanced packaging, HBM availability, rack-scale networking and data-center power deployment, which broadens the earnings-revision opportunity to TSM, MU and AVGO rather than making this solely an NVDA event. If supply unlocks, Nvidia’s revenue upside will be most visible in backlog conversion and shipment timing; if it does not, elevated demand can coexist with quarterly delivery slippage.

Over the next 1-3 months, the key catalyst is a guidance raise supported by confirmed supply allocations and sustained gross margin, not commentary on end demand. A guide that remains below the implied capacity opportunity would likely compress NVDA’s near-term multiple, since the market already capitalizes a high conversion rate from AI demand into revenue. Watch customer concentration and hyperscaler capex: a delay in cloud deployments due to power/interconnect bottlenecks would defer accelerator revenue while leaving suppliers with less direct exposure to final system acceptance relatively better protected.

The contrarian view is that the incremental profit pool may migrate down the stack as Nvidia’s scarcity premium normalizes. Greater chip availability can reduce urgency-driven pricing and increase buyer leverage, while HBM and advanced-packaging suppliers retain pricing power because their capacity additions are slower and less fungible. Over 6-18 months, rack-scale architecture also raises the strategic value of networking and custom silicon, favoring AVGO if customers pursue lower-cost inference alternatives; this is a structural risk to NVDA’s valuation premium, not necessarily its revenue growth.

Key Decisions for Investors

• Maintain NVDA exposure only as a defined catalyst trade into the next earnings release; add on evidence of a revenue-guide increase accompanied by stable gross-margin outlook. Reduce if management attributes upside primarily to theoretical capacity rather than contracted supply, or if gross-margin guidance falls materially despite higher revenue.
• Pair trade for a 3-6 month horizon: long MU and/or TSM versus short a smaller notional NVDA position. This expresses supply-chain bottleneck monetization while reducing single-name expectations risk; exit if HBM pricing softens, TSM advanced-packaging capacity expands faster than expected, or NVDA demonstrates sustained margin expansion.
• Add AVGO on pullbacks as a complementary AI-infrastructure exposure over 6-18 months, contingent on evidence that rack-scale networking attach rates and custom-silicon demand are accelerating. The principal risk is a hyperscaler capex pause, which would hit networking and ASIC programs alongside GPU demand.
• Set an event watchlist for hyperscaler capex guidance, HBM supplier commentary, TSM CoWoS utilization and data-center power/interconnect delays. A synchronized deterioration in any two of these indicators would falsify the near-term supply-unlock thesis and argues for cutting AI-semiconductor beta rather than buying the dip.