Nvidia's CEO Says There's No AI Bubble: Here's What the Numbers Say

The piece argues that recent volatility and overvaluation fears in tech mask strong underlying fundamentals: the Nasdaq Composite has been essentially flat since October 2025 (23,348 to 23,461, <0.5% gain), the Nasdaq‑100 average P/E is 32.9 (down from 33.4 a year ago), Nvidia reported ~65% profit growth last quarter with a P/E ~47.7, Microsoft’s profits rose ~60% YoY despite a 10% post‑earnings drop, and Alphabet posted >$100 billion in quarterly revenue with profits up ~33% (offset by a $3.45B antitrust fine). Jensen Huang framed multi‑hundred‑billion‑dollar secular shifts—CPU to GPU migrations, generative/agentic AI use cases, and a driverless‑car product unveiling—as justifying continued AI-driven growth, and the author concludes current valuations do not mirror dot‑com extremes, presenting a potential multi‑year buying opportunity rather than a classic bubble collapse.

Analysis

Market structure: The primary winners are GPU designers/suppliers (NVDA and partners), hyperscalers (GOOGL, MSFT) and SaaS firms monetizing generative AI (META ad lift cited). Losers are legacy CPU-centric vendors (INTC) and network incumbents with limited AI roadmap (CSCO, ORCL) as pricing power shifts to GPU and data-center compute services; expect multi-hundred-billion incremental cloud spend over 3–5 years and sustained tight GPU supply to support elevated ASPs for 12–24 months.

Risk assessment: Tail risks include tightened US/China export controls on advanced accelerators, broad antitrust actions (Alphabet, Meta) and a sentiment-driven drawdown like MSFT’s 10% post-earnings reaction; a meaningful catalyst could erase 20–40% of index gains in weeks. Immediate (days) — earnings-driven volatility; short-term (0–6 months) — mean reversion and supply shocks; long-term (1–5 years) — secular reallocation to GPU/cloud if regulatory/commercial barriers don’t emerge. Hidden dependency: revenue growth tied to hyperscaler CAPEX cycles and energy/electricity cost inflation for data centers.

Trade implications: Favor concentrated, size-controlled long exposure to NVDA and GOOGL via defined-risk structures (6–12 month call spreads) and hedge with index puts; underweight INTC/CSCO via short or put spreads. Consider pair trades: long GOOGL or MSFT vs short INTC to capture secular GPU share shift; harvest premium on highly bid IV names via short-dated iron condors where you have directional hedges. Entry: scale into longs on intra-day/weekly pullbacks >8–12% over next 4–8 weeks; trim after any 25% realized upside.

Contrarian angles: Consensus underestimates regulatory/export risk and overestimates a dot-com parallel — profitability today materially reduces bankruptcy tail but not valuation compression risk. Mispricing exists in legacy semis (INTC) where downside is not priced for GPU displacement; conversely NVDA’s leadership premium could be overdue for mean reversion if supply normalizes or AI monetization slows. Unintended consequences: faster AI rollouts raise power demand → upward pressure on energy prices and potential higher real yields, which would re-rate growth multiples negatively.

Key Decisions for Investors

• Establish a 1–1.5% portfolio notional long in NVDA via a 6–12 month call spread sized to be ~20–25% OTM (defined-risk) to capture continued GPU-led upside while capping drawdown.
• Initiate a 2% long position in GOOGL (or 1.5% long GOOGL + 0.5% MSFT) funded by a 1% short position in INTC (buy 3–6 month put spread) to express secular margin shift from CPUs to GPUs.
• Open a 0.75–1% short of CSCO via a 3-month put spread or small outright short if Cisco rallies >10% from current levels; target |-15%| to |-30%| move as a payoff window on weak AI transition execution.
• Buy a 0.5–1% portfolio hedge: 3–6 month QQQ puts ~5% OTM (or equivalent index put spread) to protect against a 20–40% sentiment shock from regulatory/export control headlines.
• Monitor BIS export-control announcements and EU/US antitrust filings over next 30 days; if new advanced-accelerator export restrictions or a major hyperscaler antitrust ruling occurs, reduce semiconductor/AI hardware exposure by 50% within 48 hours.