NVIDIA has announced that co-founder and sole CEO Jensen Huang, who led the company for 33 years, has resigned effective immediately and stepped down from the board. CFO Colette Kress will serve as interim CEO, while Synopsys veteran Harvey Jones becomes the company's first board chairman. The permanent CEO search will consider both internal and external candidates.
The news broke at 8:48 AM Eastern Time on the 12th. By the pre-market open, NVIDIA shares had fallen about 10% to around $156, breaking below the 52-week low of $164.27. Roughly $420 billion in market value evaporated in a single day, and the company's market capitalization has shrunk by about one-third over the past month.
That evening, research notes from 14 Wall Street banks flooded the inboxes of analysts and fund managers.
Three firms downgraded NVIDIA: Bernstein, KeyBanc, and William Blair. Six firms cut their price targets: Bernstein, JPMorgan, RBC, Cantor, Goldman Sachs, and Citi. But the vast majority maintained buy ratings — the core logic being that "orders are booked through 2028, and fundamentals are safe in the near term." The average price target across 13 banks sits at roughly $260, implying about 50% upside from pre-market levels.
On the question of who could succeed Huang, Seaport's Jay Goldberg wrote in his research note — "In short, No One Can Fill Jensen's Leather Jacket." Sixty direct reports, and virtually zero successors. Bernstein analyst Stacy Rasgon lamented, "Where did he find the time?"
The real concern is that the AI ecosystem financing system Huang personally built will suddenly come into question once he departs.
Goldman Sachs: Departure Is Significant but Manageable — Like Apple After Jobs
Goldman Sachs chose to wait until the official press release was issued before putting pen to paper. Analyst James Schneider characterized Huang's departure as a "meaningful but manageable negative event."
He maintained a buy rating but cut his price target from $300 to $220. His valuation logic is straightforward: EPS forecasts remain untouched at $10.00, but the valuation multiple was compressed from 30x to 22x, citing reduced visibility and elevated execution risk during the management transition.
In one sentence, the sell-side consensus: E (earnings) unchanged, P/E (valuation multiple) sharply lower.
Schneider compared the event to Apple losing Steve Jobs. Over roughly 18 months around 2011, Apple's forward P/E slid from about 17x to 12x, bottoming near 9x in 2013, only recovering to around 15x once the market confirmed the iPhone machine hadn't stopped running without Jobs.
Schneider believes NVIDIA will undergo the same "search for a reasonable multiple" process until leadership and business momentum regain clarity.
But Schneider himself acknowledged the analogy isn't perfect — Apple had a pre-selected successor, no DOJ investigation, and no congressional subpoena threats.
Bernstein's Longtime Bull Turns: "Where Did He Find the Time?"
Bernstein's Stacy Rasgon has been one of Wall Street's most respected NVIDIA bulls for years. This time, he downgraded the stock to market perform, slashing his price target from $400 to $200. The research note's title was a single sentence: "Where did he find the time?"
He wrote:
The new allegations lead us to believe the board could no longer justify keeping him in this position (we've seen other CEOs fired for far less...). If the board doesn't act now, the subsequent risk only grows, as more allegations appear likely to surface.
He also used a metaphor that will likely be the most memorable line for NVIDIA holders this week:
If you find two ants in your kitchen, the best estimate of the total number of ants in your kitchen is absolutely not two...
The implication: more allegations, more investigations, what a new leader might discover upon taking over — uncertainty is far from bottoming out.
"Sold Out Through 2028" — The Bulls' Core Argument
Bernstein, KeyBanc, and William Blair chose to downgrade NVIDIA; Bernstein, JPMorgan, RBC, Cantor, Goldman Sachs, and Citi cut their price targets by amounts ranging from $60 to $200.
But the average price target across 13 banks still sits at roughly $260 (median $250). Wedbush and BNP Paribas both have price targets of $345, more than double the pre-market price.
Most firms still maintain buy ratings. The only sell rating comes from Seaport's Jay Goldberg with a $180 target — ironically, that price is still above the current market level.
So why does almost everyone remain bullish?
The bull case boils down to one point: Huang left, but the order book stayed.
Piper Sandler's David O'Connor (overweight, $300 target) put it most bluntly:
Platform and architecture decisions for the next two generations of products are already done. The company's 2027 orders are completely sold out, and 2028 is largely the same. What's being discussed with customers now is 2029 and beyond.
Wolfe's Chris Caso (outperform, $275 target) was even more direct:
If the CEO was asked to step down, we believe fundamentals and earnings will be completely unaffected at least through 2028 — because NVIDIA is essentially sold out during this period. The CEO transition will have no impact on orders or even the product roadmap, which is clearly planned through 2028 and beyond.
RBC estimates that current demand exceeds supply by about 50%. JPMorgan notes that the production ramp paths for Vera Rubin, Rubin Ultra, and Feynman are already clear.
In other words, E is fine — the only thing moving is P/E. And there's not much multiple left to compress: at Wednesday's close, NVIDIA's forward P/E on 2027 estimates was about 10x, which Piper called a 20-year low and William Blair called an all-time low (just 7x on 2028 consensus earnings).
"No One Can Fill Jensen's Leather Jacket": 60 Direct Reports, Zero Successors
If the next two years are safe, what is Wall Street really worried about?
One major concern is organizational structure. Seaport's Goldberg summed it up in one line:
Simply put, no one can fill Jensen's leather jacket.
He acknowledged that many executives could run NVIDIA, but none carries the same weight across the entire industry. The problem is both personal and structural:
This siloed structure means any internal candidate would be deeply entrenched in their own area of expertise while knowing little about most of the rest of the business. This won't be a permanent obstacle, but it will take many quarters to reorganize the company around a new leader. Beyond that, this will be a different company — capable, but far less dynamic than it is now.
Wells Fargo put a number on it: about 60 executives report directly to Huang. Analyst Quatrochi said outright that this structure is "unsustainable and unreplicable," potentially impacting coordination speed and time-to-market.
As for the successor shortlist, the most-mentioned internal name is Ian Buck — a 22-year NVIDIA veteran who oversees hyperscale and high-performance computing and is widely regarded as the architect of CUDA, described by Piper as "well-regarded by investors." William Blair added Jonathan Ross (Groq founder and TPU architect, who joined NVIDIA only last December). On the external side, media reports mention Apple's Tim Cook, AMD's Lisa Su, Arm's Rene Haas, and Micron's Sanjay Mehrotra.
JPMorgan split the question into short-term and long-term dimensions:
We believe the most important near-term question for the stock is whether NVIDIA's board can successfully assemble an interim management team capable of maintaining the company's strong execution and advancing the existing roadmap. However, beyond the next 2-3 years, the core focus shifts to whether the management team has the forward vision to identify the next inflection point in accelerated computing and build not just chips, but a complete ecosystem and platform.
In other words: an interim team might still be able to ship the Rubin product line; but the ability to foresee "the next CUDA" a decade before everyone else is what truly made this a $5 trillion company.
Wedbush reviewed historical precedents: Marvell (which ultimately thrived under an external CEO after the founding family exited), Apple (where Cook performed well under an existing succession plan), and HP and Intel (both of which lost CEOs abruptly due to scandals, had no ready successors, and subsequently fell into what Bryson called "catastrophic periods").
But Wedbush concluded that no case fully matches the current situation:
In this case, uncertainty could intensify further, because we cannot think of any historical precedent that simultaneously satisfies all three of the following conditions: a founder-CEO with comprehensive influence over company direction, a company (and its CEO) so pivotal to technology and tech policy, and virtually no obvious succession plan at the time of transition.
NVIDIA Is AI's "Central Bank," and That Bank Has Only One Credit Officer
Wall Street has another weighty concern: what about the money?
NVIDIA is not just a chip company. For years, Huang channeled NVIDIA's free cash flow — roughly $50-60 billion per quarter — into the entire AI ecosystem: equity investments in OpenAI, Anthropic, and xAI; asset backing for emerging cloud companies like CoreWeave; prepayments and equity stakes in optical component suppliers Lumentum, Coherent, and Corning.
UBS's Tim Arcuri listed this as the number one of three most impactful areas:
NVIDIA has become a critical source of funding for numerous AI industry participants, including emerging cloud service providers, large language model developers, ecosystem partners, and suppliers. We have always viewed Huang as the central force driving this effort. Even if a new leader ultimately maintains a similar approach, there is likely to be a "pause period"... We estimate NVIDIA generates about $50-60 billion in free cash flow per quarter, a significant portion of which is recycled back into the entire system. This could easily withdraw $50 billion or more from the system — precisely at a critical moment when interest rates are rising and liquidity is tightening in parts of the credit markets.
William Blair's conclusion was more direct, and it was the core reason for its downgrade:
Existing contracts will continue to be honored after a CEO change, but what's truly at risk is the next round of deals — and NVIDIA's ability to defend market share in 2027 and beyond. Without Huang, or with him severely constrained by investigations and scrutiny, we believe NVIDIA could open a wider gap for competitors like Broadcom and AMD.
Seaport's Goldberg went further, noting that a $500 billion financing platform plan Huang announced two months ago, involving six private equity funds, has yet to have its contracts finalized. He believes that without Huang, this plan will likely dissipate.
On the day the news broke, CoreWeave fell more than 11% pre-market — on the very same day it was marketing an investment-grade loan of over $4 billion to the market.
Government Investigations Don't Disappear With the Resignation
Huang's resignation does not mean regulatory pressure is lifted.
The board's independent investigation led by Wachtell's Joon Kim is still ongoing, the DOJ's Megaspeed investigation continues, the House has requested a closed-door briefing on November 18, and Senators Warren and Coons have demanded related records.
Two banks provided specific financial quantification:
Goldman Sachs noted that NVIDIA's FY26 annual report shows total sales exposure to mainland China, Hong Kong, Taiwan, and regions outside the U.S. at about 11%, and expects increased scrutiny for operations in jurisdictions like Singapore.
UBS believes stricter rules could affect shipments to Singapore, Malaysia, and elsewhere, estimating about 10% of revenue at risk.
RBC noted that China accounted for about 1% of data center revenue last quarter and has already been excluded from guidance.
Wolfe took a different view, arguing that export concerns are overblown since AI chip shipments to China were already minimal.
NVIDIA's Pain, Competitors' Opportunity
Nearly all firms agree on one thing: competitors benefit.
BNP Paribas, Seaport, and William Blair all named AMD and Broadcom as the most direct beneficiaries, as hyperscalers and AI labs have more reason to diversify their chip sources. BNP Paribas projects that by 2030, custom chips could account for nearly three-quarters of AI accelerator shipments.
William Blair also pointed to a deeper logic: CUDA's moat is essentially built on the high cost of porting code to other chips. Now, AI coding assistants are rapidly driving down that porting cost. Anthropic mentioned in its IPO roadshow materials that getting a new accelerator up and running used to take 6 to 12 months, while its Claude Mythos Preview model reportedly achieved prompt processing on AMD's MI355X in about three days.
Analyst Naji cautiously noted that this case is fairly limited. But the fact that such news circulated in the same week the founder departed is itself telling.
Everyone Points to the Same Bottom
So where is the stock's bottom?
Wall Street's answer is surprisingly uniform — and based on experience, such uniformity is itself not a good sign.
Cantor believes 9x 2027 estimated EPS, corresponding to about $150, is a reasonable bottom. Piper identifies $150 as support at 10x trough-cycle valuation. KeyBanc's bear case is also $150. Citi agrees, corresponding to 11x valuation.
The pre-market price of $156 leaves only a 4% buffer from Wall Street's consensus bottom. The only lower target comes from RBC, whose downside scenario is $100, corresponding to 6x 2027 estimated EPS, below NVIDIA's five-year historical valuation low.
Goldman Sachs declined to give a bottom number, using that Apple chart to remind the market: trough-cycle valuations can persist at low levels for years.
Comments