Nvidia vs. Broadcom: Comparing Recent Quarterly Revenue Trajectories

Nvidia's quarterly revenue rose from $35.1 billion in Q3 2024 to $96.2 billion in Q2 2026, versus Broadcom's increase from $14.1 billion to $29.6 billion, widening Nvidia's top-line lead. Broadcom reported 86% year-over-year revenue growth and guided for $34.8 billion in fiscal Q4, implying 93% growth, while Nvidia forecast about $108 billion in Q3 revenue and 70% growth next fiscal year despite supply constraints. Both companies' 54%-66% operating margins and accelerating sales reinforce continued AI infrastructure demand, though semiconductor cyclicality and potential eventual demand normalization remain risks.

Analysis

The relevant distinction is not relative revenue scale but monetization durability. NVDA's higher operating leverage makes each incremental data-center dollar disproportionately accretive, but also leaves the equity more exposed to any utilization-driven pause in hyperscaler capex; AVGO's infrastructure-software base provides a lower-beta cash-flow offset. AVGO is also the cleaner beneficiary if large customers increasingly diversify away from merchant GPUs toward custom ASICs, where its design, networking and packaging relationships can capture AI spend without requiring a wholesale reduction in total AI capex.

Near term (days to one month), the market will likely reward upward guidance, but the more investable signal is whether lead times, supply allocation and customer prepayments remain elevated. Over the next 1-3 months, watch hyperscaler capex guidance, HBM availability, CoWoS capacity and AI-cluster networking attach rates: constraints can preserve NVDA pricing while simultaneously pushing customers to AVGO custom silicon and Ethernet fabrics. A 6-18 month risk is that software and model efficiency reduce compute per unit of output faster than inference demand expands; that would compress the market's terminal-growth assumptions well before reported revenue declines.

Consensus treats NVDA and AVGO as parallel AI-beta exposures, understating their different failure modes. NVDA is primarily exposed to a concentrated accelerator procurement cycle and regulatory/platform scrutiny, while AVGO is exposed to execution on custom programs, customer concentration, integration debt and a potentially slower enterprise-software renewal environment. The more contrarian setup is therefore relative: AVGO can outperform in a healthy AI market if spending broadens from training accelerators into networking and customer-specific silicon, and can also be relatively insulated if GPU supply rather than end demand is the immediate bottleneck.

Key Decisions for Investors

• Initiate a 3-6 month long AVGO / short NVDA dollar-neutral pair only after confirming the next hyperscaler capex updates remain intact: it targets AI-spend broadening and custom-ASIC/networking share gains rather than requiring a downturn. Exit if NVDA's next-quarter data-center guide accelerates materially while AVGO's AI segment or custom-program commentary disappoints.
• Maintain NVDA as the higher-conviction directional AI exposure into the next earnings cycle, but size below benchmark-risk limits and take partial profits following a post-results gap if forward supply commentary weakens. Thesis is falsified by a sequential data-center guide miss, a meaningful decline in gross/operating-margin outlook, or evidence that customer deployments are slipping rather than supply constrained.
• Use a basket hedge against AI capex deceleration rather than a broad semiconductor short: long AVGO versus short SOXX for 3-6 months is preferable if AVGO's software cash flows and custom-silicon backlog remain resilient. Cover the hedge if memory, advanced-packaging and optical-component lead times re-extend, signaling that physical cluster deployment remains supply—not demand—limited.
• Set alerts around major cloud-provider earnings for capex growth, depreciation-life assumptions and disclosed AI revenue. A synchronized moderation in capex plans or a rise in unutilized AI capacity would be the earliest actionable warning for both names, likely preceding estimate cuts by one to two quarters.