(Adnkronos) – Having one of the most coveted jobs in the world, that is, being a member of the board of directors of Nvidia, the company that has been the most capitalized ever, has suddenly become an unspeakable nuisance. Jensen Huang, the CEO, is too big to fire, too important to be fired, and at the same time a walking scandal that with every passing hour opens a crack in the empire that holds up the entire artificial intelligence sector.
He, 63, founder and undisputed dominus of the semiconductor giant, has admitted to an affair with a woman who, according to American intelligence, is a Chinese spy. She, Alice Huang, an executive at Megaspeed, a Singaporean company under investigation for alleged violations of American restrictions on technology exports to China. The Information has just reported that the chips smuggled into China would be worth at least 7.5 billion dollars.
Huang apologized to his wife, children, and employees. The board of directors has opened an investigation, while professing confidence in the founder. Donald Trump, who has made the Taiwanese one of his favorite advisors on AI, defended him, while confirming that the FBI is handling the matter. Wall Street was less lenient. Yesterday Nvidia lost 4.3%, burning another 190 billion dollars. Shareholders have already filed a first class action.
Jensen’s story, as his friends and even aspiring friends call him, perfectly told by Alessandro Aresu in his book Geopolitics of Artificial Intelligence (Feltrinelli, 2024), is much better than a novel, and Nvidia is the most successful, and most extreme, experiment of contemporary technological capitalism.
Huang was born in Taiwan, also lived in Thailand, and in 1972 his parents decided to send him to the United States with his older brother. He was 9 years old and ended up in Kentucky, at the Oneida Baptist Institute, which the family imagined was a prestigious boarding school. In reality, it was a facility that also housed adolescents with serious disciplinary problems.
Upon his arrival, Jensen found himself in a room with a seventeen-year-old who showed him the scars from knife wounds received during fights. His roommate couldn’t read: Huang taught him how, and in return learned to lift weights. He started doing a hundred push-ups every night. The small, thin Asian boy became a target for bullies. Not the fairy tale of Silicon Valley prodigies, but that of an immigrant who grew up amidst difficulties, humble jobs, and an almost obsessive determination. “No one cleaned bathrooms better than me,” at Denny’s, the diner chain where he worked while earning his electrical engineering degree at Oregon State University, where he met Lori, his future wife. After his master’s at Stanford, in 1993, at thirty, he found himself at a Denny’s with two fellow engineers, Chris Malachowsky and Curtis Priem, to discuss a new venture. The initial idea was to develop graphics chips for video games. 33 years later, those semiconductors would become the engine of the AI economy.
Chris Miller, in the fundamental “Chip War”, explains what distinguished Huang from all others: it’s not the quality or speed of the chips, but having understood much earlier than rivals that computer graphics cards could serve a more important purpose than making a video game realistic. In 2006, Nvidia introduced Cuda, software that allows programmers to use GPUs for applications other than graphics. It was an almost crazy gamble: according to an estimate by the company itself cited by Miller, by 2017 Nvidia had invested at least 10 billion dollars in software development. Huang distributed it for free, with one small detail: it only worked with Nvidia chips.
In 2012, AlexNet arrived, a neural network trained using two Nvidia GPUs, demonstrating the potential of accelerated computing for machine learning. In 2022, ChatGpt caused an explosion in demand for computational power. Nvidia found itself selling the infrastructure needed to build the future.
Meanwhile, the stock has grown almost thirteen-fold. The company was the first to exceed the 5 trillion dollar capitalization threshold, before the recent drops linked to the midterm elections and the Chinese honey trap. According to analysts’ projections, revenue could exceed 400 billion this year, compared to less than 27 billion four years ago.
Larry Ellison, founder of Oracle, recounts that he and Elon Musk had to beg Huang to sell them more processors. But it wasn’t a matter of price, there simply weren’t enough for everyone.
His black leather jacket has become as recognizable a uniform as Steve Jobs’ turtleneck, and Nvidia conferences resemble rock concerts (though as a speaker he is more soporific than Apple’s founder). In Taiwan, he is surrounded for autographs and photographs. One of his jackets was auctioned for a million dollars.
Alongside Huang are less famous figures like Bill Dally, Nvidia’s scientific mind, and Morris Chang, founder of Taiwan’s TSMC, which produces the most advanced semiconductors. But Nvidia’s power depends on a perfect combination of infrastructure, production chains, energy, capital, and political power. Chips are designed in the United States, manufactured mainly in Taiwan, assembled into complex systems, and installed in data centers distributed worldwide.
This is what Aresu calls “political capitalism”: a system in which corporate strategies and national security interests become increasingly difficult to separate. This mixture, which has transformed Huang from an entrepreneur into a whisperer to American presidents, now risks being fatal to him.
In June 2024, Jensen was in Taipei for Computex, one of the major events in the global technology industry.
One evening, in a bar, Alice Huang told some executives that she could convince Nvidia’s founder to join them. She sent him a message. Shortly after, Jensen arrived, ordered a whiskey, and socialized with those present. A photograph shows them sitting close together. At the time, Megaspeed’s success was just one of many pieces in the growth of the Nvidia ecosystem.
The company, born from a spin-off of the Chinese 7Road group, operates from Singapore and offers computing capacity through data centers in Southeast Asia. Its clients include Chinese companies, including Alibaba.
But in the autumn of 2025, American authorities began to ask increasingly serious questions about the destination of the processors purchased by the company.
A report published in September 2026 by C4ads, a research center specializing in the analysis of opaque commercial and financial networks, reconstructs the system through which advanced chips can reach Chinese users by circumventing American restrictions. Megaspeed is in the document: according to the data examined, it imported Nvidia hardware worth 4.6 billion dollars between 2022 and 2025, through a proprietary structure with possible ties to China.
Megaspeed denies having violated export rules. Beijing rejects accusations that Alice Huang is a spy. And there is no evidence, for now, that Jensen knowingly facilitated an intelligence operation.
But as Michelle Nie and Daniel Remler of the Center for a New American Security (Cnas) observe, Chinese companies can gain access to advanced processors through cloud services located in third countries, even without physically transferring those chips to China. You can prevent a computer from crossing a border, but not its computing power. And in any case, there are plenty of ways to circumvent controls: a Georgetown University estimate speaks of over 450,000 advanced chips illegally introduced into China in 2024 and 2025.
While the American government tried to impose restrictions on the most advanced chips, Huang asked to ease export restrictions. His argument, not unfounded, is that selling American technology to the Chinese benefits the United States. Better a China dependent on the Nvidia ecosystem than a China forced to build a national alternative (because sooner or later it will find one, as we have seen).
The CSIS also observed that Western controls have accelerated Beijing’s push for self-sufficiency in semiconductors.
But there is the opposite argument: allowing China to acquire advanced technology means helping it develop AI capabilities that could have military and intelligence applications.
Now the main proponent of the first position finds himself embroiled in a scandal that offers formidable arguments to the proponents of the second.
Huang had become a personal interlocutor for Donald Trump and influenced his choices on AI. Nvidia was a symbol of American economic power and Huang a protagonist of technological diplomacy, from relations with Gulf countries to negotiations with China.
The slip into Alice Huang’s arms changes everything.
According to reports from the New York Times, Nvidia’s general counsel, Tim Teter, had contacted a senior official at the Commerce Department while an investigation into Megaspeed was underway. The administrative inquiry was halted, and the Justice Department took over the investigation. There is no smoking gun, we don’t know if Nvidia caused the halt or exerted illegitimate pressure, but in some cases, the sequence of events is enough.
The scandal, perhaps not by chance, comes at the politically worst time. After the midterm elections, Huang risks being caught between Democrats, intent on strengthening controls over large technology companies, and Republican “hawks” on China, who criticize his permissive export policy.
Three House committees, led by Republicans, have requested a closed-door hearing. Elizabeth Warren and other parliamentarians are urging stricter measures on semiconductors. Huang could retain leadership of Nvidia and lose the most valuable part of his power, namely his credibility as an interlocutor for the White House.
At the Santa Clara headquarters, Huang works at a huge oval table. His collaborators sit in front of him, with other employees behind them. Too crowded to be a top manager’s office, it’s more of an operations room managing over sixty direct reports. He asks managers for emails with the five most important things they are working on. He intervenes in strategic decisions and operational details, down to employee salaries. There is no recognized number two. And a successor has never been discussed.
Here we return to the terror in the eyes of that board member.
Nvidia does not have a board chairman. Instead, it has a lead independent director, Stephen Neal, former head of the Cooley law firm. According to the company, this figure is a tool for independent oversight. But several board members have worked with Huang for decades, some since before Nvidia became the giant it is today. In short: the governance has remained in the days of graphics card nerds, while the company has exploded in size and complexity.
Huang only owns 3.6% of the shares. He does not control the majority stake nor has he equipped himself with special voting rights that would make him untouchable, like Zuckerberg or Musk.
So yes, the board can fire him. But who could take his place?
American media have already made a list of names: Colette Kress, CFO, is one of the few executives known to Wall Street. Debora Shoquist has extensive experience in operational management and the supplier chain. Bill Dally is a fundamental scientific figure. However, no one combines the technological authority, industrial vision, political relations, and commercial charisma of the founder.
When Steve Jobs definitively left the company in 2011, Tim Cook had spent years preparing for the succession. Nvidia has not built a similar path.
Neither of Huang’s children, Madison and Spencer, both in important positions within the company, seem to represent a credible short-term solution today. The founder stated that, should they aspire to lead Nvidia, they would have to earn that role.
We are facing the most extreme version of key-man risk: the risk that a company’s value depends excessively on a single person. Brilliant founders are rewarded by shareholders because they don’t resemble normal managers, but then when it comes to replacing them, everything risks collapsing.
The most probable solution, at least in the very first days, is that Huang remains.
The board has already declared confidence in the founder. Elon Musk defended him, Trump continues to support him, and numerous investors believe that his sudden departure would cause far greater damage than the scandal.
A first scenario is one of supervised permanence: Huang retains the role of CEO, while Nvidia strengthens board oversight, controls on conflicts of interest, export compliance, and especially the succession plan. This would be a profound transformation for a company built around the autonomy of its founder.
The second scenario is more traumatic. If the investigation reveals information withheld from the board, undisclosed conflicts of interest, or concrete problems for national security, Huang’s position could become untenable. In this case, Nvidia would have to resort to temporary leadership, perhaps collegial, while seeking a successor capable of combining technology, finance, and geopolitics.
Finally, there is a third scenario: Huang remains, but Nvidia changes anyway.
Because the company’s future does not depend solely on the founder’s capabilities. It also depends on the duration of the AI investment boom, competition from AMD and chips developed internally by large cloud operators, the ability to provide energy to data centers, and the willingness of governments to allow the spread of its technologies.
Nvidia is also no longer just a supplier to AI protagonists. It invests directly in companies that are or will be its customers, helping to finance an ecosystem that in turn buys its products. A network capable of enormously accelerating growth, but which makes the financial sustainability of the AI race even more fragile.
Huang’s political capital has been one of this system’s great assets. If that capital deteriorates, the entire sector pays the price.
In an interview with Bloomberg, Huang explained that he did not want to become chairman of the board to allow other board members to evaluate him and, if necessary, fire him. “Unlike many CEOs, I can absolutely be fired.” If that was true, it will soon be discovered.