Nvidia Would Have to Add Almost a Whole Broadcom to Reach $7 Trillion

Nvidia would need to add roughly $1.55 trillion in market value, or rise about 28% from approximately $225 to $290 per share, to reach a $7 trillion valuation. The company reported fiscal Q2 2027 revenue of $96.2 billion (+106% YoY) and net income of $59.7 billion (+126%), while guiding fiscal Q3 revenue to $108 billion and forecasting roughly 70% revenue growth in fiscal 2028. A $7 trillion valuation could be supported by about $700 billion in fiscal 2028 revenue and an approximately 18x forward P/E, though continued multiple compression remains the principal risk.

Analysis

The key debate is no longer whether NVDA can grow earnings, but whether incremental AI spend remains sufficiently concentrated to sustain its premium economics. Supply relief is a mixed blessing: it converts backlog into revenue near term, but also gives hyperscalers greater leverage over pricing, accelerates deployment of AMD systems and internally designed ASICs, and reduces the scarcity premium embedded in NVDA's gross margin. The most important second-order readthrough is HBM availability and advanced-packaging lead times; easing in either would be positive for revenue recognition but negative for the assumption that margins can remain structurally insulated.

Over the next 1-3 months, consensus revisions and hyperscaler capex commentary should matter more than another strong NVDA print. A stock already priced on distant earnings needs both estimate durability and a stable-to-expanding multiple; even modest evidence that cloud customers are shifting marginal workloads to custom silicon can compress the multiple despite upward EPS revisions. AVGO is the cleaner competitive hedge because custom AI accelerators monetize the same customer capex pool, while its software cash flows reduce dependence on one hardware cycle.

The contrarian view is that the market may be underpricing the duration of inference demand but overpricing NVDA's share of that demand. Training clusters favor the full-stack CUDA ecosystem, whereas inference becomes more cost-sensitive and workload-specific as models mature; that is where ASICs, lower-cost accelerators, networking optimization, and customer-designed silicon gain relevance over 6-18 months. The thesis is falsified positively if NVDA maintains pricing and gross-margin resilience while customer concentration falls; it is falsified negatively by a material reduction in hyperscaler capex guidance, HBM/packaging normalization without corresponding volume upside, or evidence of custom-silicon deployments displacing GPU purchases.

Key Decisions for Investors

• Maintain a tactical long NVDA only into the next earnings/revision cycle, sized smaller than a core semiconductor position; target a 10-15% upside over 1-3 months if estimates rise, but exit on a gross-margin guide below 70% or a material sequential slowdown in data-center demand.
• Initiate a 6-12 month pair trade: long AVGO / short NVDA in equal dollar amounts. This expresses AI-capex durability while hedging the risk that inference and hyperscaler custom silicon capture incremental spend; reassess if NVDA's accelerator revenue continues to materially outgrow AVGO's AI semiconductor segment for two consecutive quarters.
• Use SMH or SOXX downside protection rather than outright shorting NVDA ahead of macro-sensitive capex updates: buy 3-6 month put spreads funded by selling lower-strike puts. The relevant catalyst is any major cloud provider cutting 2027 capex plans, which would pressure the entire AI hardware complex rather than NVDA alone.
• Monitor HBM suppliers and advanced-packaging capacity as an alert, not a trade recommendation: a sharp improvement in supply lead times without an offsetting increase in system demand would signal margin and share-risk for NVDA before it becomes visible in reported revenue.