Odd Lots · Nov 14, 2026 · 36 min
Nvidia after Jensen Huang: an analyst on leadership, China policy and risk
Episode: How Wall Street Is Bracing For the Post-Jensen Huang Nvidia Era · With Stacy Rasgon
TL;DR
Bernstein analyst Stacy Rasgon downgraded Nvidia after Jensen Huang's dismissal, arguing that key-person risk, China policy and an uncertain roadmap cloud the outlook ahead of earnings.
Key takeaways
- Rasgon downgraded Nvidia after reported relationships with subordinates undercut trust in Huang, making his dismissal much more likely in his view.5:32
- Rasgon expects near-term business to hold up, with customers not expected to buy less data center hardware next year because of the leadership turmoil.7:04
- Nvidia now reports revenue by customer headquarters location rather than shipment destination, which changes how Southeast Asian sales appear.10:15
- Direct data center shipments into China are essentially nil, largely because Chinese authorities are not allowing purchases even where US licenses exist.12:47
- Huang's flat, direct-control management sped up decisions, but many deals rest on handshakes that Rasgon thinks a new CEO will struggle to replicate.15:38
- The bigger medium-term concern is future architecture and memory decisions, since Nvidia's dominance rests on bets Huang made years earlier.17:26
- Customers are spreading orders across suppliers, so the key question is whether the overall market grows large enough for several vendors.26:31
- Most of Nvidia's large cash outlays go toward securing memory, wafers and supply rather than backstops and financing of customers.30:07
Summary
Hosts Joe Weisenthal and Tracy Alloway recorded on Friday, November 13, ahead of Nvidia’s earnings the following week. They framed the leadership change as a case of key-person risk, citing reports that Jensen Huang, who built the company, was intensely involved in its business and had no clear succession plan. They also flagged legal questions and the company’s role in financing much of its customer ecosystem. Guest Stacy Rasgon, a managing director and senior analyst at Bernstein Research who covers semiconductors, called the situation a disaster with no good outcomes. Rasgon explained his downgrade of the stock. He had initially thought Huang could stay, since shareholders did not want him gone, but disclosures about relationships with subordinates, which he said had been misrepresented to the board, changed that view. Dismissal was confirmed the next day. His team spent more time deciding how to respond than writing the downgrade, which moved the stock from outperform and halved the price target without changing earnings estimates. He still sees the stock as cheap and near-term business as not much at risk, but expects more investigations and damaging headlines, which makes ownership hard to advocate. He suggested the board may have erred either in letting Huang go or in keeping him. On geography, Rasgon said Nvidia now reports revenue by customer headquarters rather than shipment destination, which can make Southeast Asian sales look different. On that basis Malaysia and Singapore are roughly 1 percent of revenue and China including Hong Kong about 8 percent last quarter, mostly gaming and auto. Direct data center shipments to China are essentially nil, largely because Chinese authorities are not letting buyers take them even where US licenses exist. He doubted that any smuggling, if it occurs, reaches 20 percent of revenue. He also said Washington may curb Chinese access to cloud compute outside China. Rasgon largely backs the company’s stance on selling older chips, finds it inconsistent to restrict the latest chips while permitting remote access, and thinks that option is now too late. Rasgon described Huang as chief technologist, strategist, top salesman and storyteller, running a flat organization where most key decisions are his. That speed was an advantage, but many deals rest on personal handshakes, which he thinks a successor will find hard to replicate. He is less worried about operations or next year’s revenue than about medium-term calls on architecture and memory across the roadmap. He credited Huang’s earlier bets on the CUDA software ecosystem, made when many doubted them, and noted that the historical share price, adjusted for a 40-for-one split, was far lower. Rasgon expects clean numbers at earnings and no material near-term change, but says management must show the trajectory holds. The interim CEO, formerly the CFO, will face questions on how the change unfolded and on keeping staff, many of whom hold large paper fortunes. On competition, customers spread orders across suppliers: OpenAI works with Nvidia, Broadcom and AMD, and Google uses its own TPUs. Rasgon said the key question is whether the overall market grows, citing an outlook where Nvidia’s guided growth of at least 70 percent points toward nearly $700 billion in revenue, Broadcom’s data center revenue roughly doubles to about $115 billion, and AMD GPU sales may nearly triple. CUDA remains a moat, though AI tools may ease porting software. On financing, the hosts raised the idea of Nvidia as a central bank of AI. Rasgon said Huang was more bullish on committing capital, but most large cash outlays go to securing supply rather than backstops. The hosts speculated that wealthy employees might reconsider their careers and that China hawkishness may rise after a public dispute within the administration over chip policy. Rasgon declined to comment on a forthcoming Anthropic filing that discusses using multiple chip types.
Quotes
“You can't sleep with your subordinates and then lie to the board about it.”
“…we spent more time with me and my team discussing what to do versus writing the actual downgrade.”
“So I actually think the ownership and who's running these things is important.”
“It's going to cause problems, I think, once he's gone, for sure.”
Who should listen
Investors and tech-industry watchers following Nvidia's leadership change and China chip policy, especially those who want a sell-side analyst's view before earnings.
Listen
This summary was written with AI from the episode transcript and reviewed before publishing. Quotes are checked word for word against the transcript, and timestamps are approximate. Quotes without a named speaker are attributed to the show.