NVIDIA: Why the Stock Fell Despite Beating Estimates — Wiki | SOPHIE

Overview

NVIDIA delivered a “beat and raise” Q3 FY26 quarter — yet the stock sold off. At 30x forward sales, the market is “pricing for perfection”: merely beating consensus isn't enough when the magnitude of the beat must keep accelerating, and it didn't. The buy-side “whisper number” ( $57B) was effectively met exactly ($ 57.01B actual), making a headline beat feel like an in-line print to aggressive funds already positioned for more.

Key Concepts

Pricing for Perfection — at extreme valuation multiples, a technical beat against published consensus isn't sufficient if it doesn't beat the unpublished, higher buy-side “whisper number.”
Guidance Deceleration — Q4 revenue guidance of $37.5B implies ~70% YoY growth, down from triple digits — the Law of Large Numbers catching up with the growth rate, not a demand problem per se.
Quick Stats (Q3 FY26) — Revenue $35.08B (+94% YoY), Data Center$ 30.8B (+112% YoY), Gross Margin 74.6% (-110 bps QoQ), Net Income $19.3B (+109% YoY).
Valuation Snapshot — Forward P/E 32.4x (below the 45x historic average), PEG 0.92 (rare sub-1.0 for a mega-cap), FCF margin 48.2% (software-like efficiency from a hardware company), Altman Z-Score 34.5 (deep in the safe zone, effectively zero bankruptcy risk).

The Bear Case: Michael Burry's Thesis

The Circular Revenue / Vendor Financing Problem — Big Tech (Microsoft, Google) invests billions into AI startups (e.g., CoreWeave), which immediately spend that capital on NVIDIA chips. Burry's argument: revenue is being manufactured by capital circulating in a loop rather than sustainable end-user demand — if VC funding dries up, the loop breaks. He draws the comparison to Cisco in 2000.
The “Air Pocket” Risk — as customers await Blackwell (B200), they may pause Hopper (H100) orders, causing a transition-period revenue dip the market reacts badly to.
CSP Indigestion — cloud providers spending $50B+/year on capex face growing pressure to show AI software ROI; if it doesn't materialize, chip orders could get cut.
Institutional Flows — notable reductions/exits: SoftBank (sold entire stake), Peter Thiel (exited near the top), Stanley Druckenmiller (reduced stake ~70%) — smart money selling into strength while retail buys the hype.
Context — despite the bearish narrative, actual short interest is only ~1.2% of float; NVDA has historically been a “widow maker” short. Burry expressed the thesis via puts (defined risk) rather than shorting stock directly.

The Bull Case: The Blackwell Supercycle

Supply Constrained — Jensen Huang confirmed Blackwell is sold out well into 2025, providing revenue visibility that acts as a floor for the stock.
Power Density Risk — reports of GB200 rack overheating requiring liquid-cooling redesigns is a real technical risk to Q4/Q1 margins, distinct from the demand-side bear case.

Retail Defense Strategy

Avoid weekly options — implied volatility is too high; premiums are overpriced relative to likely realized moves.
The Barbell Approach — hold the bulk of exposure (e.g. 80%) in broad index funds, a minority (e.g. 20%) in semiconductors — avoid going all-in on a single mega-cap name.
Watch the 50-day SMA — a break below the 50-day moving average risks triggering algorithmic selling, but also marks a technical entry zone for long-term holders.

Key Takeaways