Option A — Single-stock deep dive: the first post-Huang print is the most important single-stock event in the market this week.
The call: STAGED VALUE LONG affirmed and upgraded — half now $150–$156, half $138–$142 washout only.
Why now: $115.8B (+103% YoY) beat $109.1B, $130B Q4 guide beat $125.1B, FY28 growth raised 70% → 75% — my $115B bull bar cleared.
The disagreement: market prices a broken company at 10.8× forward; the quarter shows a working machine with a governance discount, not a demand break.
The level that changes everything: daily close below $135 kills the trade — above $180 trim half toward $200.
Last night's earnings-eve piece set two bright lines: invalidation on a Q4 revenue guide below $111B, bull case on $115B-plus with FY28 reaffirmed. NVIDIA cleared the bull bar with room to spare.
Q3 FY27 (quarter ended Oct 25): revenue $115.8B, +103% YoY, +20% QoQ vs $109.1B LSEG consensus. Adjusted EPS $2.68 vs $2.48 expected. Data-center revenue $107.6B, +110% YoY vs $101.87B StreetAccount — 93% of sales. Non-GAAP gross margin 74.9% vs ~74% expected. The company returned $28.1B in the quarter and still holds $227.3B in buyback authorization. Q4 guide: $130.0B ±3% ($126.1–$133.9B) vs $125.1B consensus. FY28 growth raised to 75% from 70%. That last number is the most important number in the report — it directly answers the pre-print fear that management would quietly walk the multi-year roadmap.
Price action tells you how much fear was priced: stock fell 24% in November into the print, six straight weekly declines (longest since 2012), closed Tuesday at its lowest since July 2025, popped ~4% after-hours, then faded to $153.26 (-1.4%) at midday Wednesday while QQQ sat at $658.38 and SPY at $730.67. The beat stopped the bleeding; it did not start a melt-up. That is exactly what a governance discount looks like.
The quarter ended Oct 25 — most of it happened before Huang resigned Nov 12. Do not read this as proof of post-Huang execution. Read it as proof that demand into the print was real.
Three facts matter. First, Vera Rubin NVL72 shipped — fastest ramp in company history, running at CoreWeave, Google Cloud, Azure, OCI and Nebius, coming soon at SpaceXSI. Kress called it "off to an explosive start." Second, backlog math: Schwab notes net backlog rose to $2.5T from $2.0T a quarter ago. Third, customers did not walk: Kress said partners are "100% behind us," TSMC visited this week, Jassy said "I don't foresee a day where we're not partners," and frontier labs (OpenAI, Anthropic) are now ~15% of data-center sales with Anthropic annualized revenue at $81B at end-October.
Leadership reality on the call: chair Harvey Jones said the board "does not intend to reinstate Jensen as CEO," Heidrick & Struggles runs the search, Kress is interim CEO + CFO, veteran Ian Buck joined Q&A (a quiet succession signal), and Bezos publicly advised "hire from internally." The 200,000-signature petition and Ives activist chatter do not change the board's stated position.
My opinion: raise FY28 to 75% is the single load-bearing commitment. Morgan Stanley had just cut calendar 2027–28 data-center estimates ~10% on Huang uncertainty and said it would "rather err on the side of too conservative." Kress went the other way and said the guide "considers data-center readiness, end-customer demand durability, and customers' access to capital." In plain English: we can build it, they want it, they can fund it.
Do the arithmetic with me. Nine-month FY27 revenue is $293.7B; add $130B Q4 midpoint = ~$424B FY27. +75% implies ~$740B FY28 revenue — above LSEG's ~$690B consensus Kress is trying to drag upward. FY28 EPS consensus $15.73 vs $9.28 this year. At $153, that is 10.8× forward earnings vs AMD at 30.4×, by far the lowest in megacap. Either the Street does not believe the 75%, or it believes it and refuses to pay for it until a permanent CEO is named. I believe the second — and that is an opportunity, not a warning, if you size it as a governance trade, not a quality compounder.
Three deductions keep this from being a clean upgrade to a full position.
First, Q4 gross margin guides to 71–72%, down ~4 points from 74.9%, on Rubin ramp. Revenue can surge while the cents kept per dollar shrink — judge this on gross-profit dollars and FCF ($43.8B in Q3), not just the top line. Opex guides to ~$11B GAAP / $10.7B non-GAAP.
Second, China is zero in the guide — again. Commerce's Monday interim rule forces license requirements for advanced chips in Singapore, Indonesia, Malaysia and Thailand, directly targeting remote-rental circumvention. DOJ/Commerce confirmed a Megaspeed probe dating to 2024; Megaspeed was once led by Alice Huang. Beijing said it "firmly opposes" and threatened countermeasures. Anyone modeling a China reopening is modeling fantasy.
Third, the financing edifice is now disclosed and enormous: $118B in guarantees (Ohio/OpenAI land/power/shell/hardware, max exposure $105B then declining over 20-year leases, zero paid to date), $102B in data-center leases/cloud commitments (up from $74B last quarter, terms to 20 years, many not yet commenced), $32B in neocloud commitments through 2032 conditional on their financing, plus a sharp jump in supply commitments tied to multi-year memory procurement. Add the Hugging Face deal (~$13B) still on track. Kress's new language — "rigorously evaluating future investments, commitments, and guarantees," "disciplined and selective" — is the right tightening, and neoclouds (CRWV +4% early, NBIS +5% early) initially cheered it. But Wednesday midday CRWV faded to $47.65 (-4.5%). The market heard discipline as conditional support. My rule stands: exit CRWV and halve NVDA longs if interim management pulls neocloud guarantees and customers cut with it.
AMD rose to $511.52 (+1.6%) and AVGO to $355.13 (+4.0%) on the sympathy bid — my AMD half-take at $580–$620 and AVGO $380–$400 targets still frame those trades. The deeper threat is not AMD share but customer silicon: Amazon will externalize Trainium in late 2027, Meta expands Vera CPU + Spectrum-X + BlueField-4 with NVIDIA even as it builds its own, Google pairs A5X Rubin instances with TPUs. And TSMC is reportedly setting credit limits / demanding cash upfront from AI customers — leverage shifting to the fab.
Why I still pay for NVIDIA: MLPerf sweeps (Training, Endpoints, Rubin inference preview), CUDA-Q quantum extensions, DSX modular factories for Europe/Middle East, and a memory/supply lock ($390M→$279B-class commitments narrative aside, direction is massive pre-buys) that a fabless rival cannot replicate in one cycle. Open-source Hugging Face + Nemotron + Palantir sovereign-AI deals widen the software trench. This is a systems monopoly for one more generation, sold at a single-manager risk discount.
S&P 500 closed Tuesday at 7,340.53, below its 200-day at ~7,347 for the first time since April — four straight down days, worst close since July. Schwab's chart note is blunt: the index broke parabolic, fell back to old resistance-as-support, a classic bearish technician's tell. VIX 23.82 (-6.5%), 10Y at 5.09%, dollar 20-month high, gold $3,787 (-2%), copper 7-month lows. Waller today floated a 50bp December hike ("shock people"); CME prices 5% for 50bp, 100% for at least 25bp. FOMC minutes drop 2pm ET today. That macro tape is why a great micro print faded by lunch.
NVDA levels: post-print after-hours high implied ~$160–$163 (Public.com prints $162.95 late), Wednesday range $151.60–$159.70. $138–$142 is the washout zone (-8.5% implied move from $154 landed almost exactly on T2 last week at $148.80). $148–$152 was T1 and held. $180–$190 is the trim wall (pre-scandal breakdown + bull-guide profit-taking). $135 is the line: a daily close below it means the market rejects even $130B + 75%, and the governance discount is becoming a demand discount.
| Trade | NVDA staged value long (101 affirmed + upgraded) |
|---|---|
| Entry | T1 $148–$152 FILLED (avg ~$150); add $150–$156 up to half size. T2 $138–$142 washout only. T3 $118–$125 panic/leadership-only. Avg ~$137 if all fill. |
| Target | $200 (trim half $180–$190 on bull-guide pop); stretch $230–$260 if insider CEO + FY28 held |
| Invalidation | Daily close below $135, or FY28 75% withdrawn, or guarantees pulled + customers cut |
| Horizon | 6–12 months (1–3 day scalp: buy $150–$154 toward $162–$168, kill $147) |
| Conviction | HIGH (beat + raise + 10.8× PE + backlog; capped by governance/margin/financing) |
| Audience / size | Half size, survivable to ~$95. Sideliners wait T2/T3. Contradiction check: pairs with QQQ short $672–$688 (hedge, not conflict) — NVDA idiosyncratic long vs index rate-risk short. |
Bull 30%: $230–$260. Triggers: Q4 $130B+ delivered with margin trough at 71% then stabilizing, FY28 75% held twice, insider CEO (Buck or internal) named by Q1, Anthropic ANT IPO (Thu Nov 19, $70–$78 range, likely above) re-rates AI capex. Neocloud financing holds, buybacks ($227B left) cushion. What it pays: multiple from 10.8× to 15–16× on $15–16 EPS.
Base 50%: $180–$210 chop then $200. Triggers: Q4 meets $126–$134B range, margins 71–72% as guided, FY28 trimmed to 65–70% on financing discipline, CEO search drags to mid-2027, China stays zero. Stock works off $135–$168 range, grinds to trim zone on execution, not re-rating. This is my central path — own it, do not chase it.
Bear 20%: $95–$125. Triggers: independent-committee findings widen, key hyperscaler cuts capex (watch MSFT already cut to $170B, META $135–$145B range), neocloud funding cracks (CRWV breaks $42 daily), 10Y >5.10% + 50bp December hike kills AI duration, S&P <7,300 forces halving per book rule. Downside is not demand evaporation — it is multiple compression on governance + financing fear.
I am wrong if any one of these prints: (1) Q4 revenue guides or pre-announces below $120B ex-China — demand, not governance, is breaking; (2) daily close below $135 post-print — market rejects the raise; (3) FY28 75% withdrawn or cut below 65% on next call; (4) top-4 customers (each 11–13% of revenue last 3 quarters; two at 18%/17% of receivables) cut orders or shift to internal silicon faster than Rubin ramps; (5) SEC/DOJ export-control action names the company, not just Megaspeed. Any single one flips me to flat; two together flip me to short toward $110.
Do not chase the morning pop. The edge is that 10.8× forward for 100%+ growth with a raised multi-year guide has never been this cheap because a person left, not because the product failed — but the quarter mostly predates his exit, so you must demand a discount for the unknown. That is the T1/T2 staging: T1 filled $148.80–$152 is a small profit at $153; sideliners wait. Add only $150–$156, never above $165. Pair with existing hedges: hold QQQ short $672–$688 toward $620–$630 (kill daily >$705), hold GLD $370–$375 toward $400 into a hawkish Fed, hold DIA/XLF/XLV rotation vs S&P <7,300 invalidation.
Second-order plays in one line each: AMD $511 hold, wait $580–$620 to trim, trail $500 weekly; AVGO $355 toward $380–$400, kill <$280; CRWV $47.65 — no second half below $54 holds daily, kill <$42; MSFT $505 hold per FCF thesis, kill on 2Q negative FCF; BTC $63,149 — WATCH-BTC-1117 intact: scalp-only $60.5–$62k if $62k holds hourly, kill daily <$59.8k, flip long only >$68k + ETF inflows; Anthropic ANT Thu — do not chase IPO day; NVDA's stake + $64B raise + $900B spend overhang will set the AI-capex tone for NVDA's T2 decision.
Acknowledging the book: 97 SHORT $180–$188 closed unfilled (no P&L, direction right); 99 FADED $75.5–$76.3k → $68k/$68k hit $64.46k closed flat; 104, 102, 101, 105 carry into today — 105's $115B bull bar is now cleared, upgrading 101 to HIGH on half size.
NVDA $153.26 (-1.42%, O $159.64 H $159.70 L $151.60 prev $155.46); QQQ $658.38 (+0.22%); SPY $730.67 (-0.24%); AMD $511.52 (+1.61%); AVGO $355.13 (+3.96%); MSFT $505.41 (-0.16%); META $614.09 (-0.77%); CRWV $47.65 (-4.49%); BTC $63,149 (-1.45% 24h); ETH $1,768.42 (-0.81%); Crypto fear & greed 11 Extreme Fear (7-day 28→11); BTC dominance 58.16%; total crypto mcap $2.17T (-4.95% 24h). Tuesday close S&P 7,340.53 (-1.03%), Dow 50,759.42 (-1.64%), Nasdaq 24,328.38 (-0.74%), 10Y 5.09%, DXY 103.76 (+0.26%), VIX 23.82, gold $3,786.70 (-2.0%), WTI $83.75 (-1.08%), BTC ~$63,900 per Schwab 9:10am ET. NVDA Q3: rev $115.844B, GM 74.8%/74.9% non-GAAP, op inc $77.5B, net $65.287B, diluted $2.70/$2.68 non-GAAP. Q4 guide $130B ±3%, GM 71–72%, opex $11.0B/$10.7B. FY tax 16–18%.
FY27E rev ~$423.7B (9M $293.7B + Q4 $130B; extra week helps Q4 ~7%). FY28 rev @75% = ~$741B vs consensus ~$690B — I model $700–$740B, EPS $15–$16 (assumes 73–74% GM blending to 72%, opex ~$44B annualized, tax 17%, ~24.1B shares with buybacks). 14× $15.50 = $217 fair value → $210–$220 range. Load-bearing input: FY28 75% growth durability. If wrong and FY28 is +50% (~$635B) on financing/China drag, EPS ~$12.50 and 12× = $150 — i.e., today's price is the bear case. Where this could be wrong: guarantees/leases convert to P&L hits; Rubin margin stays <72%; permanent outsider CEO breaks roadmap; 50bp December hike compresses all AI multiples ~15–20%.
NVIDIA Q3 FY27 press release (Nov 17, 2026) · CNBC live coverage — beat, $130B guide, board will not reinstate Huang · Schwab Market Update — Nov 18 open, backlog $2.5T, S&P below 200-day · Nasdaq — ANT IPO Nov 19 · Anthropic S-1/A-2