$NVDA · NVIDIA Corp.

Thesis

Options price a 5.4% move — under the 4-quarter average — while the stock is actually down into the print and only up 13% YTD. That's a cushioned setup for a beat-and-guide-in-line reaction. Base case: Q2 rev prints $93-94B, Q3 guide $98-100B, muted single-digit pop.

What would flip it

Q3 revenue guide below $95B would fire the guide-cut-overrides-print template and flip the reaction.

The market's narrative

AI capex bubble skepticism has capped the run: NVDA is YTD +13% (vs SPX +14%) while the rest of the AI complex (PLTR, CRWD, ORCL) trades parabolic — the crowd already thinks the bar is impossibly high and is fading the print.

Where the Street may be wrong

• Options implied move 5.4% is BELOW the 12-quarter average of 7.4% — market is UNDERPRICING volatility on the single-biggest print of the summer. Setup for a positive skew: outsized upside if guide clears $97B.
• Q1 datacenter rev $71.6B (+93% YoY) with Blackwell only 1 quarter into ramp; Q2 will be first full-Blackwell mix quarter — GM step-up + volume flywheel means the beat magnitude is asymmetric.
• China DC compute is NOT in current-quarter guide (H20 restriction still in force) — any hint of license grants or resumed shipments = discrete upside surprise vector the street is not modeling.

Peer read: AMD reported hot in early Aug (MI350 momentum) and TSM July revenue re-accelerated — the datacenter demand read-throughs are constructive, not bearish.

Reasoning

• Consensus EPS $2.09 / rev $92.07B; guide is $91B. NVDA's beat cadence on rev is $2-3B, so print $93.5-94B is the base case. My EPS $2.14 stacks 5 cents above street on GM stability + share buyback pace.
• Options implied 5.4% vs 12Q avg 7.4% = under-priced vol. Combine with hype_neutral setup (YTD +13%, not parabolic, actually down -0.97% into the print) and downside is cushioned. Payoff skew favors long.
• Q3 guide is the whole game. Street models Q3 rev ~$97B; NVDA typically guides $1-2B above, so guide range $98-100B is the base case = beat + guide fires the UP template (moderate, not extreme, because Q3 guide of $103B+ needed for the full parabolic surge template).
• Blackwell (B200) is only 1 quarter into ramp — mix shift to higher-ASP Blackwell + memory-shortage-induced 15% AI server price hikes reported in trade press = GM tailwind that street forward models have not fully baked in.
• Templates in play: forward-guide-blowout inversion (if Q3 guide >$103B → UP big); PLTR-trap DOES NOT apply this cycle because NVDA is NOT parabolic entering print (unlike CRWD/PLTR); mega-cap-with-clean-print-and-in-line-guide = default +2-4% muted UP.