[Mou Weekly] What if Jensen Huang comes back? What if he doesn't? (Week 46, 2026 | Issue No. 296)
I'm LongbridgeAI, I can summarize articles.NVIDIA fell 12.75% last week, marking its sixth consecutive weekly decline—the longest streak since 2012. In the same period, AMD rose 18.9%, Broadcom gained 18.2%, the Philadelphia Semiconductor Index climbed 9.4%, and the S&P 500 edged up 2.6%. The market hasn't abandoned AI; it's dumping the 'one-stop-shop' narrative.
On Thursday, Jensen Huang stepped down as President, CEO, and Director, with CFO Colette Kress taking over on an interim basis. This week, everyone has been asking the same question: Can he come back? While this warrants a serious answer, from a portfolio perspective, it's more critical to stress-test both scenarios.
Figure 1: Two divergent paths in the same week
A 30% probability, but not until year-end next year
Let's look at the hard pricing. On Polymarket, a contract asks when Jensen Huang will return to NVIDIA: before the end of November (3%), before the end of this year (7%), and before the end of 2027 (31%). On Thursday, the figure for the latter was still 14%, doubling after employee calls and Musk's 表态。
There are two nuances here. First, the contract scope is broad: returning isn't limited to the CEO role; serving as a director or formal advisor also counts. The 30% figure already includes the possibility of 'returning without steering,' making the probability of a CEO-level return even lower. Second, the trading volume for these contracts is only around $150,000—a single large order can move the price. This reflects sentiment direction, not a deeply considered consensus.
Even the most optimistic capital is betting on post-one-year horizons, not next week. My own estimate is slightly lower than the market price.
[Image: Polymarket 'Jensen Huang rejoins NVIDIA by…?' contract page screenshot showing prices for three timeframes]
Behind the noise: overlooked facts
The news cycle has been loud: employee petitions, Musk's signature, institutional letters to the board. But several easily ignored facts, when viewed together, carry different weight.
First, many compare this to Altman in 2023. Back then, ~700 out of ~770 OpenAI employees signed a petition, and he returned five days later. This time, among NVIDIA's 40,000+ employees, only 1,600+ comments under an internal suggestion box post have been verified. A public petition site shows 30,000+ signatures, but only ~2,500 claim to be current NVIDIA employees—self-reported and unverified. The viral 'core employee joint letter threatening mass exodus' on X remains unconfirmed; fact-checkers note its wording almost verbatim copies the 2023 OpenAI letter.
Second, the institutional letters don't demand reinstatement. Brad Gerstner (Altimeter) and Gavin Baker (Atreides) wrote to the board asking them to slow down, avoid hastily appointing a permanent CEO, and keep the option of Huang's return open. The same letter states they trust Kress and acknowledge the board could ultimately determine that his return isn't in shareholders' best interests.
Third, the nature of this crisis differs. Altman's removal had vague reasons and no specific issues were found afterward. This case has three layers: facts admitted by Huang himself in his employee letter; reported relations with a direct subordinate (a corporate governance issue, with shareholder class-action lawsuits filed); and national security concerns—U.S. intelligence officials suspect a woman closely tied to him is collecting intelligence for China. The first two layers fall under the board's purview; the third does not.
Fourth, the timeline is out of his hands. The board has hired external law firms for an independent investigation; he has agreed to testify before Congress before Thanksgiving. After the midterm elections, the new Congress taking office in January will ask even harder questions. Bringing him back before the investigation concludes would mean the board overturning itself, exposing individual directors to litigation.
Fifth, the door is open because the process is slow, not because they're waiting for him. Reports say the board contacted Arm CEO Rene Haas; the board responded that headhunters are still screening candidates and no formal outreach has occurred. The search process is just starting, but the direction is toward hiring a new leader.
Consider precedent. Large-cap CEOs who left due to relations with subordinates—Intel in 2018, McDonald's in 2019—did not return. Founder returns like Jobs, Dell, and Schultz all left for reasons unrelated to personal conduct.
Figure 2: Same 'bring him back' slogan, but different context
If he returns: stock gaps up, but the landscape won't revert
Assume the 30% scenario plays out. On the day of his reinstatement, NVIDIA will likely gap up, reversing much of the risk premium added last week.
But three things won't go back. Investigations and congressional testimony won't end with his return. Major clients have seen the 'single-source' risk; supply chain diversification has been on their agenda for three years, and this provides the strongest rationale yet—they won't drop their second supplier just because he returns. Moreover, the market has priced in 'NVIDIA = one person' for the first time; once key-person risk is visible, it won't be ignored again.
His return is an opportunity to reduce risk, not a reason to go fully long.
If he doesn't return: chips enter a Warring States period
NVIDIA's product roadmap is visible for years; chips won't degrade simply because the CEO changes. What he takes isn't technology, but the hand that aligns everyone.
This hand does four things. Allocation: Advanced nodes, advanced packaging, and HBM are scarce capacities; NVIDIA secures the largest share through relationships built deal-by-deal with upstream partners and willingness to sign commitments others won't, totaling $279 billion in long-term supply agreements. Capital backing: In H1, NVIDIA made ~$42.4 billion in equity investments; for every $1 of revenue, ~$0.30 flows back to customers and AI infrastructure, effectively using its balance sheet to stimulate demand. Pricing power: The 75% gross margin relies on customer lock-in. As HSBC's Frank Lee bluntly noted during his downgrade, a new CEO will struggle to maintain the same pricing power with customers and bargaining power with suppliers.
Talent retention: He has ~60 direct reports and has never announced a succession plan. Veteran engineers who've followed him for 10–20 years largely don't need money; they stay for him. Former NVIDIA GPU engineer Yun-Ta Tsai has publicly warned that poor handling could trigger mass departures of key technical staff. Industry sources suggest competitors are already recruiting them. While currently unconfirmed, this is the fastest-acting risk among the four.
Rumors circulating on X (industry hearsay, unverified)
None of these four factors are on the roadmap, nor will they automatically continue under interim CFO leadership.
Without Jensen Huang, NVIDIA is unlikely to dominate alone. NVIDIA captures over 80% of AI processor revenue and the bulk of profits. This share peaked in 2024 and was already softening; his departure accelerates the trend. AMD serves as the second supplier for general-purpose GPUs; Broadcom designs custom chips for major clients, generating $16.7 billion in quarterly AI chip revenue (+2x YoY); cloud providers are designing in-house; CPU leaders were never monolithic. GPUs and CPUs will likely re-enter a fragmented era, ending NVIDIA's era of 独占 majority profits. Broadcom's stock-to-NVIDIA ratio rose 35.4% last week, its largest single-week gain since IPO.
For the industry, this may be positive. Over the past three years, downstream spend enriched upstream monopolies, keeping compute expensive. Competition will lower compute unit costs, enabling model builders, cloud providers, and app developers to profit, extending the chain's longevity. The pie hasn't shrunk; there are more slices. But industry tailwinds don't equal NVIDIA shareholder tailwinds.
On the flip side: swapping stocks takes a day; swapping suppliers takes years. Software ecosystems and contracted capacity remain intact. The 'Warring States' period spans years, not quarters. AMD reversed nearly 9% in a single day on Friday; how much of last week's rally was emotional remains to be seen.
[Cartoon: The 'King' seat on the chessboard is empty; pieces from all sides move toward the center]
Two questions, one exam
Whether he returns or not, the answer lies in the same report: NVIDIA's earnings after hours on Tuesday, with Kress facing the conference call solo.
If NVIDIA's upcoming report remains strong and core client procurement shows no slowdown, this week's volatility will likely prove a temporary repricing of valuation and risk premiums.
Conversely, if earnings reveal delays in order fulfillment, shrinking guarantee scales, or decelerating client capex, Wall Street's logic will shift from a simple 'CEO transition' to a substantive reset of the AI investment cycle.
The $108 billion revenue guidance for this quarter faces little surprise. Key items to watch: Will next quarter's guidance be raised? Are the terms for customer advances and upstream contracts tightening? Is the language on client demand still 'accelerating'? Is there a talent retention plan? Options market pricing for tonight stands at ±9.6%, the highest since August 2024.
Figure 3: One earnings report, two paths
My approach
My NVIDIA position is fully long. I'll assess reducing risk exposure next week, implementing hedges before earnings.
The logic is straightforward. His return is a one-year horizon issue; position sizing is a Tuesday night issue. You can't use hope for the former to bear risk for the latter. Entering earnings fully long bets solely on the bullish scenario.
Post-earnings, two paths emerge. If Scenario 1 materializes, remove hedges and gradually reduce positions to comfortable levels based on the 'no longer sole source' premise. If Scenario 2 unfolds, reductions must extend beyond NVIDIA to the entire AI hardware chain.
NVIDIA remains a quality company, now cheap to absurdity. Based on next year's earnings expectations, P/E is ~11x, the lowest since 2016; based on the year after, it drops into single digits. Toyota trades at ~11x forward. A company with near-doubling annual revenue and a legacy automaker growing 3% annually command similar market valuations today. The S&P 500 averages 18x.
The market prices in the risk that the 80% share and 75% margin won't hold. Thus, cheap ≠ worth full position. Between 'good company' and 'worth full position' lies the 'sole source' assumption. When premises change, positions must follow. Foregoing some gains is cost; suffering extra losses is accident.
[Next Week Outlook]
This week, AI trades face two exams within 48 hours.
Last week, the broader market moved more calmly than NVIDIA; equal-weight S&P rose 2.4%, VIX closed at 19.3. Pressure stems from rates: October CPI beat expectations, Fed Governor Waller publicly suggested considering a 50bps hike in December, and the 10-year Treasury yield reclaimed 5.15%.
This week's calendar: Tuesday morning retail sales, after-hours NVIDIA; Wednesday FOMC minutes, Anthropic pricing; Thursday Anthropic IPO, Walmart earnings.
Tuesday tests the old king's ledger; Thursday tests the new 贵's pricing power.
Buy signals remain unchanged: equal-weight S&P holds support, VIX stays below 30. Both conditions persist.
Regardless of personnel changes, check the books first.
(Views expressed are personal and do not constitute investment advice.)
The copyright of this article belongs to the original author/organization.
The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.
