Why Nvidia's board still finds it hard to replace Jensen Huang after he admitted to an extramarital affair
On November 8, Jensen Huang admitted to an extramarital affair and apologized. Nvidia's board announced the launch of an independent investigation the same day, but still expressed confidence in him and his leadership team. The article argues that a flat organization in which more than 60 people report directly to Jensen Huang, the board trust accumulated from the RIVA 128 in 1997 and from the roughly ten years it took for the CUDA investment starting in 2006 to pay off, and his deep ties in communications with TSMC, top customers, and the government make the cost of replacing him too high. As of March 23, he beneficially owned about 3.58% of the shares, had no super-voting rights, and held no chairman title.
Why read this
AI score 55
Breaking down why Nvidia finds it difficult to replace its CEO from three layers—organizational structure, board trust, and internal and external relationships—can help understand the company's governance structure.
Topics
Body · AI translation
[[**After Jensen Huang admitted to marital infidelity, Nvidia's board still publicly supported him; his management authority built over more than three decades, his track record, and his internal and external relationships make replacing him too costly, and the company has little incentive to change CEOs in the short term.**
**Key points**
**1. A flat organization amplifies Huang's personal authority**
More than 60 people report directly to him, the "Top 5 email" can reach the grassroots directly, and a culture of speaking frankly makes his judgment the standard for work; a successor would need to rebuild this system of coordination and trust.
**2. Long-term victories have built board trust that is hard to hand over**
In 1997, at a moment of desperation, he bet on the RIVA 128; from 2006 he invested in CUDA for about a decade before it paid off, and later pivoted to AI and acquired Mellanox; continuously delivered returns make the board willing to tolerate high-risk investments.
**3. A low stake does not hinder his actual influence**
As of March 23, his beneficial stake was about 3.58%, and he has neither super-voting rights nor the title of chairman; but core directors have profited alongside him since the 1990s, and supporting him is equivalent to supporting the existing way of doing business.
**4. Internal and external relationships and the current state of the investigation make replacing him too costly**
Family members and "Nvidia second-generation" figures have entered different businesses of the company, and communication with top customers, TSMC, and the government is also deeply tied to him; on November 8 the board launched an independent investigation while still expressing confidence, and replacing him would disrupt operations and increase uncertainty in the handover. 2026-11-11 11:15 Nvidia can only forgive Jensen Huang Letter List Quick Read
This article comes from the WeChat public account: Letter List, author: Xiao Jinya
Jensen Huang is mired in scandal, but Nvidia finds it very hard to replace its CEO.
On November 8, Jensen Huang admitted to marital infidelity and apologized. On the same day, Nvidia's board, in a statement announcing the launch of an independent investigation, still clearly stated that it had confidence in him and his leadership team.
Behind this support are very realistic considerations of interest. For more than thirty years, Jensen Huang has repeatedly turned his own judgment into the company's direction, and used business victories to earn greater trust.
Jensen Huang shaped Nvidia's way of working, and also brought rich returns to those who followed him. The more successful the company becomes, the more this dependence is reinforced.
Replacing Jensen Huang now would mean asking a group of people who have already won many times to voluntarily give up the helmsman they know best.
Never say never, but the cost of Nvidia abandoning Jensen Huang is simply too great.
First, the most obvious point: the fewer the layers, the more important Jensen Huang is.
Nvidia's flat management reinforces Jensen Huang's personal authority. The company keeps management layers very thin, so that business lines converge directly before him.
The more problems he participates in solving, the more the entire organization becomes accustomed to relying on him to operate.
In an interview with Lex Fridman this year, Jensen Huang said that more than 60 people report directly to him, almost all from engineering fields, covering different areas such as memory, CPU, and GPU. He does not schedule regular one-on-one meetings; instead, he brings the relevant leaders together to discuss problems jointly.
This is closely tied to Nvidia's business. The company now has to combine chips, networking, memory, and cooling into a complete computing system, and every part must work together.
Bringing people together eliminates repeated relaying of messages and layer-by-layer reporting. Jensen Huang sits among experts from various fields, pressing them on proposals, coordinating disagreements, and taking part in deciding how to make trade-offs. The distance between departments has shrunk, and his role within it has become more prominent.
And his channels for obtaining information have always extended down to the grassroots.
According to Tae Kim, author of The Nvidia Way, Nvidia has a "Top 5 Email" system that has continued for many years. Every one or two weeks, employees write emails about what they are working on, or competitive trends and new technologies they have observed, and send them to their teams, managers, and email groups that Jensen Huang can access.
Jensen Huang reads hundreds every day, and when he comes across something that interests him, he will directly question the employee.
And on the basis of a highly flat structure, Nvidia has also developed a culture of Jensen Huang being "outspoken." Even project leaders may be reprimanded in front of the whole company.
At an all-hands meeting in 2011, when Jensen Huang talked about the Tegra 3 chip, he asked the photographer to point the camera at project leader Mike Rayfield and kept zooming in. After everyone could see his face clearly, Jensen Huang began criticizing the project's progress and demanded that he get the chip right.
Jensen Huang later explained that he would give a person feedback in front of others because everyone else should also learn from mistakes. He also opposed having managers listen to instructions privately and then relay them to the team, believing that important information should be heard by the relevant people at the same time.
This has embedded his management style deep into the company's daily operations. Jensen Huang's personal standards of judgment have gradually become work standards that everyone is familiar with and follows.
Therefore, replacing Jensen Huang would also mean the entire organization needs to adapt anew. A successor would have to take on a large number of specific matters and also make each department believe that he or she can make equally effective trade-offs.
The appointment of a CEO can of course take effect in an instant, but this kind of trust may take a long time to build. Jensen Huang has already spent more than thirty years building it.
Jensen Huang holds a kind of power at Nvidia that is very difficult to hand over.
He can ask the company to sacrifice immediate profits and keep investing in a market that may only mature years later. The board and employees are willing to follow him because several of his major judgments ultimately changed the company's fate.
This trust was first established at a moment of life and death.
In 1997, Nvidia had enough cash to last only about six more months. The earlier NV1 and NV2 had bet on the wrong technology path and were incompatible with Microsoft's DirectX, so the company had to develop its products all over again. Under the usual process—first build sample chips, then test, revise, and re-produce—the money would very likely have run out before the product reached the market.
Jensen Huang and his team decided to use simulation equipment to verify the design in advance and to stake their funds on mass production before they even had physical samples in hand. The RIVA 128, launched later, kept Nvidia alive.
Years later, recalling this experience, Jensen Huang admitted that the company had made many wrong decisions, but at the most critical moment, it got several things right in a row.
More than a single turnaround from the brink, the later investment in CUDA better demonstrates Jensen Huang's astonishing powers of persuasion.
In 2006, Nvidia began releasing GPUs that supported CUDA, enabling them to handle computing tasks in addition to graphics processing. This investment did not yield sufficient returns for many years, and shareholders at one point wanted the company to refocus on improving profitability.
The cost was almost unavoidable. Chip costs rose, yet the market lacked corresponding applications, and gamers were unwilling to pay more for the extra capability. Jensen Huang recalled that CUDA lowered the company's gross margin. When the host asked how long it took for this investment to start paying off, he answered, about ten years.
Today, when people talk about CUDA, the first thing they think of is Nvidia's competitive moat. Back then, however, supporting it meant accepting an investment that would hurt the profits of the existing business and that had no large-scale market for a long time. Before the board, Jensen Huang repeatedly explained why he believed in this path. The directors ultimately accepted his reasoning and allowed the company to keep investing.
By 2013, this trust helped Nvidia pivot rapidly to AI.
Greg Estes, a vice president at Nvidia, remembers that Jensen Huang once sent an email on a Friday night announcing a full pivot to deep learning, and by Monday the company was already moving in the direction of AI.
In his account, Nvidia's strategic shift was so fast that it took only a single weekend.
Jensen Huang saw the opportunity and also had enough authority to make every department adjust its work immediately—and the flat organizational structure deserves part of the credit for that.
In 2019, Nvidia announced it would spend about $6.9 billion to acquire Mellanox, filling in high-speed networking capabilities. When large numbers of processors work together on computing tasks, the efficiency with which they exchange data among themselves is crucial. The acquisition announcement already proposed that future data centers would need to be designed as a single integrated computing system.
As these decisions accumulated, Jensen Huang's weight within the company naturally grew heavier. The board and longtime employees had followed him through investment, doubt, and waiting, and had seen the returns with their own eyes. The next time he proposed a bold plan, there would only be less resistance.
Jensen Huang's influence over Nvidia far exceeds his shareholding percentage. The trust accumulated through long-term cooperation, together with the interest relationships formed by shared profits, has given him a position in the company that is very difficult to shake.
For many directors and longtime employees, supporting Jensen Huang also means supporting a way of doing business that has benefited them for years.
According to Nvidia's shareholder documents this year, as of March 23, Jensen Huang beneficially held about 3.58%.
The company implements one share, one vote; he has no super-voting rights and no chairman title. Nvidia does not have a chairman; instead, an independent lead director assumes the corresponding responsibilities.
But Jensen Huang's weight is clearly not something that can be explained by his shareholding percentage.
Nvidia's board now has 11 members, several of whom were already involved with the company back in the 1990s. More than thirty years later, they have watched with their own eyes as Jensen Huang took a small company that could have gone under at any moment to where it is today.
Mark Stevens is one of them. In 1993, he participated in Sequoia's early investment in Nvidia. At the time, Sequoia invested $1 million, and Nvidia was valued at about $7 million. He left the board in 2006 and returned two years later at Jensen Huang's invitation.
By June 2024, the Nvidia shares he held were worth about $5 billion.
Another director, Brooke Seawell, joined in 1997. He later recalled that the graphics chip market at the time was highly competitive, Jensen Huang was just over 30, and before founding the company he had not even served as a vice president. This young man, who initially made him uneasy, later led the company to grow continuously.
By June 2024, the Nvidia shares Seawell held were worth about $700 million.
These experiences affect how a director views the CEO. Outsiders see a bold bet; they look back and remember how previous bets weathered the troughs and delivered returns.
And these long-term relationships have already extended to the next generation, forming "Nvidia second-generation."
Jensen Huang's daughter Madison currently serves as senior director of product and technical marketing for Nvidia's physical AI platform, participating in promoting platforms such as Omniverse, Cosmos, and Isaac. The business she is responsible for is precisely in the direction of Nvidia's expansion into robotics and real-world applications.
His son Spencer works in the robotics business. His son-in-law Nico Caprez serves as vice president of global AI infrastructure growth, responsible for contacts with new cloud service providers and data center operators.
Jensen Huang's family members have already entered different business segments of the company.
Beyond the Huang family, co-founder Chris Malachowsky's son also works at Nvidia, and director Aarti Shah's son joined the company in February 2024.
Anthony Robbins, Nvidia's former head of public sector business, also once worked at the company with his daughter. In March 2025, when the father and daughter were interviewed together on a podcast, the father had just retired, while the daughter had been with the company for about five months, responsible for channel distribution business in the public sector.
Another employee, Elena Lewis, wrote in a public post congratulating her father on his retirement that he had worked at Nvidia for 24 years, and that she had been fortunate to work alongside him for five of those years. She said she was proud to be a "second-generation Nvidian." At the end of the post, she signed herself as "Lewis, still at Nvidia."
Jensen Huang is open-minded about employees referring their own children. He believes parents would not casually recommend candidates who might embarrass them.
Nvidia has thus formed a group with stable relationships and mutual familiarity. Jensen Huang sits at the center of these relationships, and his personal experience has long been tightly intertwined with the company's growth.
Removing Jensen Huang would add a business risk to Nvidia. The relationships and credit he has built outside the company are likewise difficult to hand over to a successor through a single appointment.
The needs of important customers, cooperation with core suppliers, and the government's attitude toward chip sales all revolve closely around Jensen Huang as the core.
In September 2024, Larry Ellison described a rather interesting dinner at Oracle's financial analyst meeting.
He and Musk had once dined with Jensen Huang at the Nobu restaurant in Palo Alto, lobbying him for more GPUs. Ellison even used the word "begging" to describe the scene at the time, hoping the other party would take more money and sell more chips.
Whether GPUs can arrive on time affects how customers' AI projects proceed. The founders of top-tier customers negotiating directly with him shows that these important deals are already closely tied to his personal work.
On the supply chain side, there are likewise long-accumulated relationships.
TSMC founder Morris Chang once said that the connection between the two began with a letter around 1997. Jensen Huang at the time hoped TSMC would manufacture for Nvidia, candidly acknowledged the company's financial difficulties, and at the same time promised that once the product succeeded, Nvidia would become an important customer of TSMC. Later, that promise was fulfilled.
The two sides also tackled production challenges together. In 2024, the Blackwell chip encountered yield problems due to design defects, affecting production. Jensen Huang stepped forward to acknowledge that the responsibility lay entirely with Nvidia, and thanked TSMC for helping fix the problem and restore production. When the delivery of important products is blocked, he bears the work of explaining the situation to the outside world and maintaining cooperative relationships.
These collaboration experiences have built up credit for him. To put it plainly, for Nvidia's many suppliers, Jensen Huang's promises carry weight.
In addition, Nvidia's business is increasingly affected by policy, and the CEO also has to directly participate in communications about market access.
In July 2025, Nvidia officially disclosed that Jensen Huang traveled to Washington to meet with Trump. The same month, the company announced that it was applying to resume H20 sales and had received assurances that the United States would issue licenses.
Therefore, what the board needs to consider is who can simultaneously take on these internal and external affairs of the company, and what price will be paid during the transition period.
On November 8, the Nvidia board announced the launch of an independent investigation, while clearly stating that it has confidence in Jensen Huang and his leadership team. Trump subsequently also said he believed his explanation, and mentioned that the FBI was investigating.
At present, there is no sign that any relevant party is seriously considering "abandoning" Jensen Huang.
For the board, a change of leader must help solve the company's problems. Right now, removing Jensen Huang would mean first disrupting a set of operating arrangements that have been running for many years, and then bearing the uncertainty of the transition.
Has published 796 articles on Huxiu. This content is authorized by the author for publication; the views represent only the author and do not represent Huxiu's position. If you have objections or complaints about this manuscript, please contact tougao@huxiu.com. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]]]