The call: Cover NVDA shorts at the open. Build a 6–12-month value long in three tranches at $148–152, $138–142 and $118–125, with a $200 target. Exit only if the Nov 17 Q4 revenue guide comes in below $111B. Use half a normal position. SPECULATIVE.
Why now: NVDA fell 8.6% after hours to $158.78 when the WSJ reported the board is preparing to oust Jensen Huang. Polymarket gives 71% odds he is out by Friday, and Wedbush puts the stock near 10x next-year earnings, against 15x for TSMC.
The disagreement: The Street says an ouster means another 20–30% down. We think removing Huang cuts the worst risk: a CEO facing a subpoena while Congress writes export law. The risk the market hasn't priced is a new CEO's audit of Nvidia's $108.5B in guarantees.
The level that changes everything: $164.27, the March 52-week low. A daily close back above it after the board's decision means the ouster is priced. If NVDA stays below, let the ladder fill.
NVDA traded at $158.78 after hours, inside our $155–160 target. The board is reportedly preparing to remove Jensen Huang, which is the clearing event we said the stock needed.
Wednesday afternoon the New York Times reported a second relationship, with Kimberly Powell, Nvidia's VP of healthcare, who reports directly to Huang. Minutes later the WSJ reported the board was preparing to force him out. CNBC confirmed the board met that night, and the meeting ran into the evening. Semafor reports some directors favor CFO Colette Kress as interim CEO. It also reports that the board briefed the White House and that an outside search firm is considering candidates including Apple's Johny Srouji. Prediction markets moved fast: Kalshi went from 37% to 64% odds that Huang is out by Nov 13, and Polymarket showed 71%. On Monday Kalshi had priced only 17% for an exit by 2027 (CNBC). The stock took three downgrades in one day. Baird cut to Neutral and its target from $500 to $200. Bernstein cut to Market Perform and its target from $400 to $200, writing: "We've had enough, and prefer to watch from the sidelines."
Losing the CEO is a known, limited cost. Keeping a CEO facing a possible subpoena was an open-ended one.
Here is what was building before Wednesday. Reps. Mast and Moolenaar said they were ready to subpoena Huang after their Nov 11 deadline passed with no reply. Sen. Kevin Cramer (R), Rep. Subramanyam (D) and Rep. Bacon (R) all called on him to resign. Warren (the incoming Senate Banking chair) and Banks (R) are pushing the AI Overwatch Act, the MATCH Act and the Chip Security Act into the lame-duck defense bill (NDAA). Subramanyam said it directly: whether Huang stays "could affect its relationship with Washington and whether lawmakers crack down." Democrats also take control of both chambers in January.
China is a smaller revenue factor than the headlines suggest. CFO Colette Kress already guided to "no China data center compute revenue in our forward outlook." Losing Huang as Nvidia's go-between with Beijing therefore costs option value, not revenue the company has guided to. Trump visits Shenzhen for APEC next week, and the US-China trade truce expires in January. A board that has removed Huang gives Washington less reason to punish the company itself.
My opinion: The bull case on governance was never that Jensen stays. It was that Washington stops escalating, and removing him is the fastest way to get there. Some exposure stays with the company whoever is CEO: Megaspeed bought more than $4B of Nvidia chips by analyst estimates (CNN), and The Information alleges about $7.5B moved to China through Alice Huang's network (unverified). That risk is in the bear case below.
At $158.78, NVDA has already absorbed about three-quarters of the "fired CEO" discount that analysts are predicting.
The leading internal names are Ian Buck, the CUDA architect who runs hyperscale and has the backing of the Yale CEO-forum authors in TIME and of Jim Cramer, and CFO Colette Kress. A Buck/Kress interim pairing named on the same day as the ouster would be the McDonald's pattern, and it is the fastest route to our target.
The real risk after Huang is not the engineering. It is that his successor may not keep making the financing commitments Huang made.
According to CNBC's tally of Nvidia's Q2 FY27 10-Q (filing):
| Exposure | Size | Why it matters after Huang |
|---|---|---|
| Equity stakes bought, 6 months | $42.4B (vs $1.2B a year earlier) | A new CEO can stop this. These stakes helped fund Nvidia's own customers. |
| Stake book on balance sheet | $98.9B | Private stakes are held at carrying value, so a new CEO could mark them down. |
| Guarantees | $108.5B ($105B of it on OpenAI's Ohio leases) | These are contingent liabilities, and they are the first thing a new CEO would review. |
| Commitment to buy back cloud capacity | $36B | Nvidia renting capacity built on its own chips is the clearest case of "circular" financing. |
| Stake gains in pretax profit | $23.7B, or 16.8% | These gains hit GAAP earnings. Non-GAAP consensus mostly excludes them. |
| Revenue from top three customers | 44% (16%, 15%, 13%) | Losing one quarter from one of these customers is roughly $17B, enough to cause a sequential revenue decline. |
JonesTrading's Mike O'Rourke made the point plainly: the successor "will not have the same latitude or deference to pursue circular deals and aggressive sales tactics." Here is the chain. If the interim team stops writing new guarantees and cloud buybacks, the neoclouds get less financing, which slows their capex and then Nvidia's orders. Baird already defers its 2027 data-center forecasts and warns hyperscale capex growth could fall to 40% or lower.
The Nov 17 report is likely to carry the first guidance issued without Huang. Interim management teams tend to guide conservatively. The options market's 6.7% implied move was priced before the ouster news, so expect it to widen. NVDA has moved by more than the implied amount in only two of its last eight reports, but one of those was a 13.8% drop. That is why our buy levels price in some write-down risk instead of buying at the current price.
At $158.78 the probability-weighted value is roughly the current price. The edge starts below $150 and is widest at $118–125, so the largest discount gets the last tranche.
| Field | DAILY-NVDA-1112 — staged value long (corrected) |
|---|---|
| Entry | One-third at $148–152, one-third at $138–142, and one-third at $118–125, the value zone. Fill tranche 3 only if the drop is driven by panic or the leadership change, not by the exit triggers below. Do not chase above $152. Average cost if all three fill is about $137. |
| Target | $200, the target Bernstein and Baird set assuming no Huang; about 12x next-year EPS. Upside of $250 or more if the multiple re-rates to TSMC's 15x. |
| Invalidation | A Q4 revenue guide below $111B on Nov 17. That would be a sequential decline that breaks the "sold out into next year" assumption. There is no price stop, because at $120 the stock is cheaper on our numbers, not wrong. |
| Timeframe | 6–12 months |
| Conviction | SPECULATIVE. Two signals support it: the valuation gap to TSMC and the board acting decisively. The confirming signal, the guidance, is still pending. |
| Sizing (the risk control) | Half a normal position. Size it so a further 30% drop, to about $95, is survivable. With no price stop, position size is what limits the loss. Long-horizon investors only. If you need a hard floor, use defined-risk options such as cash-secured $120 puts or call spreads. |
| Price | × next-year EPS (~$17) | × run-rate EPS ($10.11) | Discount to $204 base value |
|---|---|---|---|
| $158.78 (after hours) | 9.3x | 15.7x | 22% |
| $145 | 8.5x | 14.3x | 29% |
| $120 | 7.1x | 11.9x | 41% |
Insider check (30 days): There were no open-market purchases. The only filings are two 500,000-share gifts by director Tench Coxe (Oct 7 and Nov 4). The pre-earnings blackout explains the silence. The most recent insider sales came in September at $218–234 (Kress, Stevens, Teter). A Huang share sale after an ouster is a real supply risk. He holds about 70M shares directly and about 800M through trusts, and as a former CEO he would face fewer optics constraints. This is my opinion and is not modeled. If it happens, it is the kind of panic selling that tranche 3 is designed to buy.
| Scenario (to mid-February) | Probability | Trigger | NVDA |
|---|---|---|---|
| Bull | 25% | A Buck and/or Kress interim is named within days, the Q4 guide is ≥$118B, and the subpoena threat fades | $195–215 |
| Base | 45% | Huang is ousted and an interim CEO named, the guide lands at $111–118B, and the probes continue | $140–175 range (tranches 1 and 2 fill) |
| Bear | 30% | No successor for more than two weeks or Huang contests the decision; the guide is <$111B; or export charges target the company or NDAA chip bills pass | $120–135. If the cause is panic or leadership, buy tranche 3. If the cause is a guide below $111B or charges against the company, exit. |
Expected value: 0.25×$205 + 0.45×$157.5 + 0.30×$127.5 = $160.4. That is +1% from $158.78, so there is no edge at the current price. It is +10.6% from $145, +17% from a $137 average cost, and +34% from $120.
The trade of buying Nvidia's competitors is now consensus, so take some profit on it and keep the neoclouds on a short leash.
Builds on our Nov 9 and Nov 10 deep dives. Upcoming dates: jobless claims Thursday; PPI Friday Nov 13; NVDA earnings after the close Tuesday Nov 17; House briefing deadline and FOMC minutes Wednesday Nov 18; Anthropic begins trading Thursday Nov 19; Trump at APEC in Shenzhen next week.
A. Data snapshot · B. Model and assumptions · C. Open-call reconciliation · D. Sources
| Item | Level | Time |
|---|---|---|
| NVDA | Close $173.73 (+2.87%; H $177.54, L $170.77). After-hours last $158.78 (−8.61%). Hyperliquid perp low $156.14 | Close Nov 11; after hours ~8pm ET |
| NVDA fundamentals | TTM P/E 21.1; TTM EPS $7.91; gross margin 74.67%; beta 2.36; 52-week $164.27 (Mar 30) to $243.37 (Oct 6) | Finnhub, 05:32 UTC Nov 12 |
| NVDA Q3 FY27 consensus | EPS $2.53; revenue $111.36B; reports after the close Tue Nov 17; implied move 6.7% (pre-ouster) | Finnhub; Investing.com Nov 10 |
| AMD / AVGO | $552.90 (+7.32%) / $345.22 (+3.99%); after hours +4.3% / +2.4% | Close Nov 11; CNN 8pm ET |
| SMH / SOXX / TSM / GOOGL | $506.96 (+4.82%) / $155.40 (+6.01%) / $413.35 (+4.25%) / $293.45 (+2.11%) | Close Nov 11 |
| SPY / QQQ | $759.81 (+1.32%) / $690.38 (+2.31%) | Close Nov 11 |
| Futures | Nasdaq-100 −0.97%, S&P −0.37%, Dow +0.15% | CNBC ~05:28 UTC Nov 12 |
| Ouster odds | Kalshi 64% (by Nov 13, from 37%); Polymarket 67% (by Nov 12) and 71% (by Nov 13) | The Block, 7:50pm ET Nov 11 |
Valuation: Annualized run-rate EPS = $2.53 × 4 = $10.11, which is 15.7x at $158.78. On next-year EPS we rely on Wedbush's "closer to 10x" at about $170. That implies about $17 of next-year EPS, which we did not compute ourselves. Base value = 12x × $17 ≈ $204, in line with the sell-side ex-Huang targets of $200. Finnhub's 33.9x "forward P/E" does not reconcile with consensus EPS and is not used.
Scenario EV: We take the midpoint of each scenario's range, weighted 25/45/30, which gives $160.4.
Why the exit is a fundamental trigger, not a price stop: In a value thesis, price falling toward intrinsic-value support makes the stock more attractive, unless the fall reflects new information that lowers intrinsic value. A price stop cannot tell the two apart. The fundamental trigger, a Q4 guide below $111B, tests the load-bearing input directly.
Load-bearing input: The assumption that demand is contracted into next year (Goldberg: "They're sold out. It's all contracted into next year"), so a change of CEO does not dent FY28 revenue. If this is wrong, for example if customers use the transition to shift orders to TPU or AMD while hyperscale capex growth falls toward Baird's 40%, next-year EPS falls about 20% to roughly $13.5. Then fair value at 12x is about $162. That is still above the $137 average cost, but the margin of safety is gone, so we exit on the guide signal instead of averaging down.