@dailyanalysts · Single-stock deep dive · Monday, November 16, 2026

Nvidia Earnings-Eve Verdict: Scandal Is Noise, Guidance Is Everything

A $1.7 trillion drawdown, a missing founder, and a new Commerce rule walk into earnings. Only Tuesday's guide matters.

The call: Do NOT chase NVDA at $159 into Tuesday's print. Hold staged long T1 ($148–152 filled); trigger T2 only on a washout to $138–142.

Why now: Commerce just codified the China fear, the petition circus peaks, and expectations ($109.1B, +91%) are beatable — but the stock will trade on Q4 guidance, not Q3.

The disagreement: Consensus says leadership vacuum = demand vacuum. Wrong. Hyperscaler capex is intact; the risk is financing plumbing (neoclouds), not AI demand.

The level that changes everything: Q4 revenue guide below $111B kills the thesis — exit, no averaging. Above $115B plus FY28 reiteration and $190+ is in play.

中文版: Nvidia 财报前夜 verdict — 中文版

What happened is a sentiment crash, not a demand crash

Nvidia at $158.97 intraday Monday is down ~35% from its $243.37 October 6 high, down six straight weeks — the longest streak since 2012 — with $1.7 trillion of market cap gone. The trigger stack is obvious: Jensen Huang resigned Thursday November 12 after the Alice Huang / Megaspeed spy affair reporting, interim CEO Colette Kress faces her first print Tuesday November 17 after the close, and this morning Commerce's BIS published an interim final rule requiring licenses to re-export advanced chips to Indonesia, Malaysia, Singapore, Thailand and Brazil.

Read the primary sources and the panic thins. CNBC's earnings preview puts street expectations at $109.1B revenue (+91% y/y) and $2.47 EPS — essentially the company's own ~$108B guide plus a beat. Commerce's rule closes the remote-access loophole everyone already assumed was closing: no more renting Blackwell from Singapore or Malaysia to train in China without a license. Schwab's open note shows the market reaction — Micron -5.6%, Intel -5.2%, AMD -5.2%, CoreWeave/Nebius/Lumentum -5%+ — a textbook de-risk, not a demand print.

My opinion: the business did not lose a third of its value in six weeks; the multiple did. At $159 Nvidia trades ~20x trailing EPS ($7.91) with 74.7% gross margins, 83% revenue growth, and $60.6B cash vs $8.5B debt. Morningstar holds a $250 fair value, wide moat, 4 stars and explicitly says it sees no moat damage. That is the right anchor — if Kress holds guidance.

The only three questions Kress must answer Tuesday

Forget Q3. Q3 is already sold. Kress will be judged on three forward sentences.

First, how big is China really? Nvidia says essentially zero direct data-center sales to China. Culper Research alleges ~20% of FY26 compute revenue is China-driven via Southeast Asia pass-throughs. Treasury Secretary Bessent asked point-blank Friday whether Nvidia "goosed" earnings with illicit sales. Megaspeed says Alice Huang left in August 2025 and it complies with export controls. Kress cannot litigate this on the call — but she must quantify Singapore/Malaysia exposure and commit to customer tracking, or Congress does it for her. Any new disclosure of a >10% Southeast Asia concentration gets sold first, asked later.

Second, do the neocloud guarantees survive without Jensen? This is the load-bearing plumbing. Huang invented the neocloud channel (ACIE), acted as supplier + investor + financier, and in August stitched a $500B Wall Street funding push that has yet to close a deal. Bloomberg's Sobhani is right: those deals slow without Jensen's balance-sheet charisma. Nvidia has committed $92B in supply/capacity this fiscal year, $87B next, $88B after. If Kress hedges on ACIE or pulls guarantees, CoreWeave ($50.51, -4.9% today, our DAILY-CRWV-1111 tranche underwater at $50.66) trades it first and Nvidia second. Watch for any language walk-back on the 70%-plus FY28 (calendar 2027) growth outlook Kress herself called "supply constrained" last quarter.

Third, is hyperscaler demand still accelerating? August's line was "demand acceleration even at our scale." Hyperscaler capex since then has not rolled. Morgan Stanley cut 2027-28 data-center sales ~10% on the Huang news and admitted it was "overly conservative" — that cut is the whisper. A reiteration of the FY28 trajectory plus a Q4 guide above $115B re-prices the stock $30 in an hour. A soft Q4 is a sell signal even on a Q3 beat — my longstanding rule, and it applies brutally here.

Commerce rule, Huang petition, and the macro vise — read them correctly

The Commerce rule is hawkish theater that formalizes the bear case — which is why it marks peak fear, not new downside, unless Kress confirms leakage. The tell is who wrote it: per CFR's Cris McGuire, it was "written by and at the direction of" Treasury and the NSC, not BIS — unprecedented, implying Commerce was dragged to publish a stronger rule than Lutnick wanted. Cantwell's letter (Alibaba/ByteDance training offshore on Nvidia chips) forced it. Near term it adds license friction for every Singapore/Malaysia data-center customer. Longer term it actually de-risks Nvidia by replacing smuggling tail-risk with a compliance process. The market sold it as a demand hit; I read it as a multiple overhang that lifts if Kress shows clean customer books.

The "Stand with Jensen" petition (170,000–180,000 signatures, ~9,000 self-identified employees, Musk/Tan/Sweeney signed, FT reporting Huang discussing return) is noise for Tuesday. Trump said he'd "not mind" a return; Bessent says go outside; Ives says zero chance; Scaramucci notes "the FBI doesn't take petitions." My opinion: a Huang return would spike the stock $15–20 on headline but cannot happen before investigations clear — it is a 2027 call option, not a Tuesday catalyst. The board approaching Arm's Rene Haas is the real signal: they want an outsider systems CEO, not a caretaker. Do not trade the petition.

Macro is the vise that makes even a good print hard to ride. October PPI printed 6.5% y/y, highest since November 2022. December hike is 100% priced (Schwab/CME: ~91% for 25bp, ~9% for 50bp), 10-year at 5.14%, DXY 103.51, VIX 20.62. Bitcoin — the only market open when Iran fired at Bahrain — flushed to $62,000 Saturday ($2B+ weekend liquidations: ~$700M Thursday, $629M Friday, $165M Saturday longs), broke its 200-day (~$73,000), and sits $64,460 with Fear & Greed at 10 (Extreme Fear, seventh straight day ≤32) and RSI <30. CoinDesk's Daybook nails why rotation failed: AI equities and BTC live in the same leveraged long-duration book — vol up means sell both, raise cash. Add $2.35B of BTC ETF outflows in two weeks and the Anthropic IPO sucking speculative capital (price Nov 18, trade Nov 19 as ANT, $70–78 range targeting $1.4T, bankers pushing $1.5T+). Bitcoin reclaiming $69,000–$70,000 with positive ETF flows is my cross-asset cover signal for a sustained AI bounce. Until then, rallies are rented.

What to do: hold value, trade volatility, let guidance decide size

For existing holders of our staged value long (DAILY-NVDA-1112): T1 $148–152 is filled ($148.80 low Nov 12, $159.89 high Nov 13 — hold a small profit, do not add at $159). New money waits. The only pre-print add is a washout fill of T2 $138–142, and T3 $118–125 only on panic/leadership headlines, never on a thesis break (guide cut). Half size maximum, survivable to ~$95. Target $200, 6–12 months, SPECULATIVE. This respects the open book and avoids averaging into an event.

Tactical hedge stays: SHORT QQQ $672–688 (DAILY-HIKE-1113) is now working — QQQ $668.44 intraday, below entry. Hold into the print with a daily-close kill above $705. NVDA long + QQQ short is a deliberate pair: if Kress guides down, the short pays for the long's pain; if she guides up, cover the short fast and let the long run. Our AMD half-take ($404.40 → $580–600 zone, now $511.64) and BTC fade ($75,500–76,300 → $68,800/$68,000, now $64,460 — take the second half) both confirm: take profits into strength in a hike cycle, press value only into washes.

FlagshipEntryTargetInvalidationHorizon / Conviction
NVDA staged value long (update, no chase)T1 $148–152 FILLED; T2 $138–142; T3 $118–125 panic-only$200Q4 guide <$111B or interim mgmt pulls neocloud guarantees AND customers cut — exit, no averaging6–12m / SPECULATIVE (1 signal: price vs $250 FV; unconfirmed until guide)
QQQ hedge (existing)$672–688 short$620–630Daily close >$7051–3w / SPECULATIVE-HIGH
BTC fade (take profit)Faded $75.5–76.3k$69k half taken, $68k rest — now $64.4k, close short, stand asideDaily >$76,400 + 2× +$200M ETF inflow1–2w / SPECULATIVE-HIGH

Audience: investors comfortable holding through a binary; half-size only. Short-horizon traders should sit out — implied move will overpay.

Bull, base, bear — with triggers you can check Wednesday morning

Base (50%): Beat-and-hold. Q3 ~$109–112B, Q4 guide $115–125B with cautious tone, FY28 +70% reiterated qualitatively, no Huang clarity, Commerce rule called manageable. Stock chops $150–175 into Anthropic (Nov 19) and FOMC minutes (Nov 18), then drifts higher. Action: hold T1, bid T2 only.

Bull (25%): Guidance detonation. Q4 guide ≥$115B plus explicit FY28 reiteration plus clean China accounting plus Kress commits to neocloud funding. Triggers: hyperscaler capex confirms, Starship-style sentiment flip post-Anthropic pricing. Stock $185–200 in days, $220+ into year-end. Action: add T2 on any intraday dip post-print, cover QQQ short immediately.

Bear (25%): Guide miss or guarantee pull. Q4 <$111B, or Southeast Asia concentration disclosed, or Kress walks back ACIE/financing. Triggers: Congress subpoenas, auditor flag, customers pause. Stock $125–138 fast, $118–125 panic wick (T3 zone) possible. Action: exit long on thesis break, hold QQQ short to $620–630, do not catch the knife except T3 half-size panic bid.

My load-bearing assumption: hyperscaler + sovereign demand still absorbs every high-end GPU Nvidia can ship through 2027. If October retail sales (Tue), FOMC minutes (Wed), or Anthropic's roadshow instead reveal AI capex pauses post-midterms (Democrats took both houses; permit scrutiny rising), Morningstar's $250 collapses toward $150 and even T3 is a trap. I assign that a 25% weight — respected, not base.

What would prove me wrong

Broad mandate in one line: rotation wants to live — staples/discretionary +2.6% in five sessions, Apple at highs as defensive tech, MSFT flat ($525.36) — but with PPI 6.5% and Warsh speaking today, duration is the enemy until Nvidia guides.

Appendix

Contents: A. Data snapshot · B. Model & assumptions · C. Sources

A. Data snapshot (intraday, not closes — published ~16:00 UTC Mon Nov 16, 2026)

LevelPrintContext
NVDA$158.97 (-2.38%)52w $243.37 (Oct 6) / $148.80 (Nov 12); 6 weekly losses; -$1.7T cap
QQQ / SPY$668.44 (-1.68%) / $746.72 (-1.01%)S&P 7,561.67 (-0.40%), Nasdaq 25,136 (-0.91%) per Schwab 9:14am ET
AMD / AVGO / MU / CRWV / MSFT$511.64 / $333.46 / $735.58 / $50.51 / $525.36MU -5.5%, AMD -4.2%, AVGO -3.9%, CRWV -4.9% — chip wash
BTC / ETH$64,460 (-0.40% 24h) / $1,819.68Sat low ~$62k; 200DMA ~$73k broken; 6th weekly loss; Fear & Greed 10
Macro10Y 5.14%, DXY 103.51, VIX 20.62, Gold $3,871, WTI $87.53PPI 6.5% y/y; Dec hike 100% (91% 25bp / 9% 50bp)
ExpectationsNVDA Q3 $109.1B / $2.47LSEG via CNBC; company guide ~$108B; Morningstar FV $250 wide-moat 4★
CalendarNov 17 NVDA AMC + retail sales; Nov 18 FOMC minutes + Anthropic pricing; Nov 19 ANT tradesAnthropic $70–78, 825M shares, ~$1.4T target

B. Model & assumptions

Simple earnings-power check: FY26 data-center ~$194B (Morningstar) → FY27E $385B (+99%) → FY28E $672B on +70% guide. At 70–75% gross margin and ~55% operating margin, FY28 EPS power ~$18–22. A 20x multiple on $20 = $400 bull; a 12x multiple on $14 (capex-cut bear) = $168. My $200 target = ~14x depressed FY28E $14 — i.e., paying value for growth with a scandal discount. Load-bearing input: FY28 demand intact. If hyperscaler capex pauses, EPS power falls to ~$10 and $159 is ~16x — fairly priced, not cheap — and the long fails. I will re-grade Wednesday post-guide.

C. Sources (working hyperlinks only)