AI News

Jensen Huang Resigns as NVIDIA CEO: How an Early IPO and the Absence of Dual-Class Shares Laid the Groundwork

36Kr · AIUpdated Nov 15 22:16

AI summary

On November 12, NVIDIA announced that co-founder Jensen Huang has resigned as President, CEO, and director, effective immediately; CFO Colette Kress will serve as interim CEO, and the board has appointed Harvey Jones as its first Chairman and launched a search for a new CEO. The article notes that when NVIDIA went public in 1999 it adopted a one-share-one-vote structure, and Huang's stake fell from about 13.1% before the IPO to 3.58% in 2026, lacking the protection of dual-class super-voting rights; his 33-year leadership rested mainly on performance-based authority rather than institutional control.

Why read this

SelectedAI score 85

The article uses Jensen Huang's resignation to trace NVIDIA's governance trajectory from its 1999 IPO through continuous equity dilution, and compares it with the dual-class share structures of Google and Meta.

Topics

Tags

Body · AI translation

Author | Huang Yi

Editor | Zhang Fan

On November 12, Nvidia suddenly announced that company co-founder Jensen Huang has resigned from the positions of president, chief executive officer, and director, effective immediately. Current Chief Financial Officer Colette Kress will serve as interim CEO, and the board has simultaneously appointed Harvey Jones as the company's first chairman in its history and launched the process of selecting a new CEO.

Since founding Nvidia in 1993, Jensen Huang's 33-year tenure at the helm has officially come to an end. From GPUs to artificial intelligence, Jensen Huang led Nvidia through several major strategic transformations, turning a startup chip company into one of the most influential technology giants in the world.

However, less than five months earlier, Jensen Huang had won about 99% of the votes in the director election at Nvidia's annual shareholders meeting. Such a stark contrast has also prompted the market to re-examine a corporate governance issue long overshadowed by Nvidia's dazzling performance:

As the company's founder and long-term leader, why did Jensen Huang not, like Google's Page and Brin or Meta's Zuckerberg, consolidate his own control through a dual-class share structure?

01 1999 Was Not Yet the Era of Dual-Class Shares

The answer to this question may need to start with the equity arrangements when Nvidia went public in 1999. At that time, Silicon Valley technology companies going public commonly adopted "one share, one vote" arrangements. For founders, they could influence company decisions through shareholding, executive positions, personal reputation, and other aspects, but they did not have voting rights inherently superior to those of ordinary shareholders. Nvidia also followed this tradition and issued only common stock.

As for the dual-class share structure, it was in fact not an invention of the internet era. As early as the last century, many well-known U.S.-listed companies used dual-class share structures to maintain control through shares with different voting rights, typical examples being Ford Motor and The New York Times. However, explicitly using this system to protect the long-term strategy of technology companies and gradually making it a commonly adopted equity structure for IPOs of large technology and internet companies was a trend that formed only later in the internet era.

In 2004, Google adopted an AB share structure when it went public. In terms of voting rights, Page and Brin set the Class A shares held by the public at one vote per share and the Class B shares held by founders and others at ten votes per share. Google's prospectus explicitly stated that a traditional publicly listed company structure could weaken corporate independence, and therefore it hoped to use AB shares to maintain management's long-term control over the company and safeguard the implementation of long-term strategy.

Figure: Historical evolution of unequal voting rights among technology stocks; Source: 36Kr chart

By comparison, Nvidia's listing was only five years earlier. Its continued use of a one-share, one-vote equity arrangement largely reflected the equity governance conventions of technology companies at that time. Moreover, Nasdaq Rule 5640 restricts listed companies from unfairly weakening the voting rights of existing ordinary shareholders after the fact through means such as issuing super-voting shares. Therefore, for Nvidia, which was already listed, the institutional difficulty of introducing super-voting rights to protect the founder was already far higher than before an IPO.

02 Jensen Huang's Stake Has Been Continuously Diluted

Nvidia's ownership is highly dispersed, and this was already evident at the time of its listing. Its 1999 prospectus disclosed that Jensen Huang held about 13.1% of the shares before the IPO, which fell to 10.8% after the IPO; by 2026, his shareholding ratio had further dropped to 3.58%. Financing, additional share issuances, equity incentives, and changes in personal holdings caused Jensen Huang's shareholding ratio to continue declining, and his voting rights weakened accordingly, laying the groundwork for this "ouster incident."

Figure: Timeline of Jensen Huang's equity dilution; Source: 36Kr chart

In earlier years, there were already precedents of Silicon Valley technology company founders losing control due to equity dilution. In 1985, Jobs and Apple CEO Sculley clashed fiercely over the company's operations. Although Jobs was still Apple's largest individual shareholder at the time, his stake had fallen to about 11%, no longer enough to sway the board's decisions. The board chose to support Sculley, and through a business restructuring, Jobs was stripped of his management authority and left the Apple he had founded in September of that year.

Looking at Microsoft, after its 1986 listing, Gates still held about 45% of the shares. Although his personal stake gradually declined, he still retained about 14% before stepping down as CEO in 2000. Thanks to Gates's relatively high shareholding at the time of listing, the relatively gradual process of equity dilution thereafter, and the leadership authority he accumulated over time, Gates maintained strong control over the company throughout his tenure as CEO.

By comparison, Jensen Huang held only about 11% of the shares when Nvidia went public. Compared with Gates, the voting rights corresponding to the shareholding ratio of the "leather-jacket warrior" were clearly weaker. Moreover, as his shareholding ratio further declined thereafter, Jensen Huang's already weak voting rights continued to diminish, meaning he had long since lost the foundation to dominate company decisions through controlling ownership.

03 Jensen Huang Led the Company for 33 Years Relying on Performance Authority

Therefore, Jensen Huang's influence over Nvidia is difficult to explain by the voting rights corresponding to his equity ratio. Management authority in corporate operations never comes only from the ballot box; it also comes from performance authority, strategic judgment, and the board's long-term trust in executives.

Moving from graphics chips to programmable GPUs, and then to sustained investment in the CUDA ecosystem, Nvidia has repeatedly bet on long-term technology roadmaps before the market reached consensus. CUDA in particular was difficult to deliver meaningful returns immediately in its early days, yet it gradually built up a competitive moat through software-hardware synergy; later, generative AI drove an explosion in computing demand, and Nvidia evolved from a chip supplier into a core participant in global AI infrastructure.

These key strategic decisions and the enormous commercial success they brought gradually established Jensen Huang's core leadership position at Nvidia. In the company's day-to-day operations, Jensen Huang has long led product R&D direction, R&D resource allocation, supply chain coordination, and customer relationship management, and the company's flat management style has strengthened his influence over frontline operations and experiential decision-making. For the board of directors and institutional investors, supporting him was both an endorsement of past achievements and an expression of trust in his future judgment.

Chart: Nvidia's long-term operating revenue performance; Source: Wind, 36Kr

Therefore, even though Jensen Huang does not hold absolute control in the legal sense, he has long possessed de facto strategic dominance, and on the strength of consistently delivered results, he has maintained his leadership position for 33 years without equity protection. In the June 2026 director election, he received about 99% of the affirmative votes, a direct reflection of this trust.

But the above "Jensen Huang model" has an inherent condition: performance authority must be continuously recognized. It can strengthen as the company succeeds, but once operating performance or personal reputation is hit, long-accumulated leadership authority may quickly weaken; in particular, Jensen Huang, lacking absolute control, may thereby lose the board's support, or even lose operational dominance over the company. This "ouster incident" is clear evidence of the above view.

04 Who Exactly Do Dual-Class Shares Protect?

Based on Nvidia's board-centered governance model, according to the company's bylaws, the board needs only the support of a majority of directors to have the right to appoint or remove senior executives such as the CEO at any time, regardless of whether just cause exists. Even as a founder, Jensen Huang cannot veto the board's personnel decisions on the strength of less than 4% shareholding alone.

For Jensen Huang, who holds a relatively low stake and lacks absolute control, even as a long-serving CEO he must rely on performance growth and shareholder trust to maintain his leadership position. Because he cannot control the board through his own voting rights, the board can form a substantive check on his power and, when necessary, push for management personnel changes. It can thus be seen that Jensen Huang's resignation this time is not surprising; it is precisely the result of the combined effect of Nvidia's long-standing equity structure and corporate governance mechanism.

Yale University professor Jeffrey Sonnenfeld noted in a recent interview that Jensen Huang successfully led Nvidia for more than 30 years without dual-class share protection, making him a powerful example against the dual-class share system; yet, ironically, if Nvidia had given him super-voting rights back then, removing him today would be far more difficult.

Google and Meta are typical dual-class structures, where super-voting rights give founders stronger control over the company, enabling them to reduce interference from short-term market fluctuations and shareholder pressure and to advance long-term strategic planning. But when a founder's judgment or behavior goes wrong, ordinary shareholders may find it difficult to change the corporate governance landscape through voting. Thus, dual-class shares amplify a founder's strategic autonomy and correspondingly reduce the convenience of external correction.

Nvidia did not adopt a dual-class structure, both because of historical factors at the time of its 1999 listing and because it formed a governance logic different from that of tech giants such as Google and Meta, namely choosing to let performance speak rather than locking down control through institutions, thereby preserving the board's ability to supervise and correct. Jensen Huang's resignation this time once again proves that no matter how great the performance authority once was, a founder lacking institutional control is always subject to the rules of corporate governance.

The content of this article represents only the author's views.

Markets carry risk, and investment requires caution. Under no circumstances do the information or opinions expressed in this article constitute investment advice to anyone. Before deciding to invest, investors should, if necessary, consult professionals and make prudent decisions. We do not intend to provide underwriting services to any party to a transaction or any services that require holding specific qualifications or licenses.

Source

36Kr · AI