Who is hunting Jensen Huang?
Original author: Sleepy
Around 5:30 PM ET on November 11, The Wall Street Journal published an exclusive report: Nvidia's board of directors was preparing to oust Jensen Huang.
Just over twenty minutes earlier, The New York Times had just exposed his second extramarital affair—with Kimberly Powell, the Nvidia healthcare business vice president who had reported directly to him for 9 years. That evening, Bloomberg followed up, reporting that the board had already been in contact with external headhunters. In after-hours trading, Nvidia fell about 9%.
As the first company in history to surpass a $5 trillion market cap, Jensen Huang had run Nvidia for a full 33 years.
Interestingly, none of the materials that were intensively released over the past two weeks in an attempt to bring down Jensen Huang involved anything new.
The photos from the first affair were taken in June 2024; the U.S. Commerce Department investigation began in September 2024; intelligence surveillance had confirmed Huang's identity back in the summer of 2025; a short-selling firm had written the relationship between the two into a 40-page public report half a year ago; and as for the San Jose hotel video in the second relationship, it was filmed in March of this year.
These materials sat quietly for anywhere from six months to two years—in secure cabinets in Washington, in law firm document pouches, on short-sellers' hard drives, and on former employees' phones—and then suddenly, over the 14 days from October 29 to November 11, they were released one after another every few days, as if on a scheduled lineup.
In Silicon Valley, which reveres Founder Mode, a founder's personal morality has always been the flaw most easily exempted. But this time, that rule failed when it came to Jensen Huang.
We wanted to figure out why—and why these 14 days in particular. In this crisis that appears to have been set off by sex scandal, who exactly needs Jensen Huang out?
A Suppressed Memo
Regarding the Singapore company Megaspeed and its managing director Huang Le, the outside world has now pieced together the full story: how a shell company spun off from a Chinese gaming company, with only $5.7 million in cash on its books, obtained Nvidia's official cloud partner license within four days, and how it bought at least $4.8 billion worth of Nvidia chips through a data center in Malaysia.
In electronic surveillance of Huang Le, U.S. intelligence agencies followed the trail and discovered her intimate relationship with Jensen Huang, and assessed that she had been recruited—with the mission of getting close to Jensen Huang and influencing Nvidia to continue lobbying Washington to relax the chip ban on China.
For more than a year, this investigation had been impossible to push forward.
The reason lay in the White House. Over the past year, Jensen Huang had become almost a regular at Mar-a-Lago and the White House.
When Trump was inaugurated in January 2025, Huang was not among the first batch of tech company founders invited, but he quickly secured his ticket.
In April, he attended a Mar-a-Lago dinner with tickets at $1 million each; subsequently, Nvidia anonymously donated $25 million to the White House State Dining Room renovation project—according to documents obtained by The Washington Post, that figure tied with Apple for first among all corporations. By this year, Nvidia had even established the first political action committee in its history, with two vice presidents overseeing software and hardware each donating $443,000 to the Republican National Committee on the same day.
The returns were immediate. In May 2025, the White House rescinded the Biden-era AI Diffusion Rule; in July, H20 chips resumed shipments to China; in December, Trump announced that Nvidia would be allowed to sell the more powerful H200 to approved Chinese customers, with the only condition being that the U.S. government take a 25% cut of sales.
On Joe Rogan's podcast late last year, Jensen Huang had recounted with some pride the opening line Commerce Secretary Howard Lutnick used when he first called him: "Jensen Huang, this is Secretary Lutnick. I just want you to know, you are a national treasure, and Nvidia is a national treasure. Whenever you need to reach the President or the government, call us directly—we're always here."
When a company's CEO is called a "national treasure" by the Commerce Secretary, the conventional enforcement machinery becomes very difficult to operate.
In August 2025, after Huang Le was interviewed by police in Singapore, Nvidia's general counsel Tim Teter called Deputy Commerce Secretary for Export Administration Kessler directly, questioning whether the U.S. side was pushing from behind. A few weeks later, frontline investigators received internal instructions that all future inquiries involving Nvidia must be approved by the deputy secretary personally, and that the name Megaspeed must no longer be mentioned externally.
According to reports from The New York Times and Axios, in the 13 months that followed, the Commerce Department's Entity List did not add a single company; the list of more than 100 companies approved by the interagency committee—including Megaspeed, which had been specifically named for inclusion—was entirely suppressed at the Commerce Department.
Even after intelligence officials reported the surveillance findings on Jensen Huang directly to Trump in October 2025, Trump decided to do nothing.
With administrative channels completely blocked, the China hawks in Washington who advocated tightening chip controls had only one other path to take.
In the spring of 2026, Secretary of State and National Security Advisor Rubio personally oversaw the drafting of a top-secret memorandum that packaged together the surveillance records, the romantic relationship between the two, and national security risks, and circulated it among Cabinet officials with high-level security clearances. On the eve of Trump's visit to China in May, Rubio even reminded Trump in person aboard Air Force One en route to Beijing.
But Trump still did not change the arrangement. He later said lightly that the matter "doesn't really affect me," and took Jensen Huang to Beijing as planned, seating him at the head table at the White House state dinner in September.
For a full five months, that top-secret memorandum went nowhere. Until late October, when someone decided to flip the table.
14 Days and the Final Blow
If you lay out the reports since October 29 in chronological order, you'll see this was not a random media expose, but a hunt precisely timed to political and business inflection points.
The first shot was fired on Thursday, October 29. The New York Times published its first investigation: "Nvidia Chief Said to Be Target of Chinese Spy Plot."
The piece unusually carried six bylines—reporters covering the White House, trade export controls, Asian geopolitics, Silicon Valley, and the intelligence beat were all on it. At The New York Times, that specification means cross-departmental verification had been ongoing for months, with sources feeding information spanning multiple levels of the White House, the intelligence community, and the Commerce Department.
Why publish on this day? Because five days later was the U.S. midterm elections on November 3, and three weeks later Trump was heading to Shenzhen for the APEC summit. Blowing this open before the summit was equivalent to preemptively blocking any possibility of Trump using relaxed chip controls as a bargaining chip.
Even more deliberate was the scale of the disclosure: the first report only said the two were close, without a word about the affair.
This was clearly a hook with a follow-up in reserve. Sure enough, Nvidia walked right into the trap that day, with a spokesperson denying it loudly, saying meeting clients was a CEO's core job and that Huang's dealings with Huang Le were "not unusual, nor surprising."
After the midterms landed and the Democrats took both the Senate and the House, the net suddenly tightened faster.
On November 5, three Republican House committee chairs sent a joint letter demanding Jensen Huang appear for a closed-door congressional briefing; the same day, in the White House Situation Room, Treasury Secretary Bessent, in front of the Defense Secretary and Secretary of State, asked Commerce Secretary Lutnick why Megaspeed still had not been added to the Entity List. The next day, Friday—the last trading day before the weekend—The New York Times and the Financial Times simultaneously dropped a second wave of reports, directly characterizing it as a "romantic relationship."
Just days earlier he had been saying "not unusual"; two days later, on Sunday, November 8, Jensen Huang was forced to send an apology letter to all 40,000-plus employees, writing the line "I was unfaithful to my wife."
Then on the following Monday and Tuesday, the focus of the disclosures shifted rapidly from private life to criminal compliance: media reported that general counsel Teter had intervened in the Commerce Department investigation last year, and that the case had been transferred to the Justice Department and converted into a criminal investigation; Bloomberg and The Information dropped trade data alleging that Nvidia had under-reported about 50,000 chips, helping a network spanning China and Singapore funnel at least $7.5 billion—roughly 240,000 restricted GPUs—into China.
But even by Tuesday night, many in Silicon Valley and on Wall Street still thought Jensen Huang could ride it out.
The reason was simple: the so-called espionage allegation was built on a memorandum that had never been made public. Beijing-based observer Fred Gao pointed out the logical flaw on Substack: Beijing had been restricting domestic Big Tech purchases of H200 this year, requiring inference compute to prioritize domestic chips, even causing 600,000 chips to pile up in Hong Kong. Why would a government that is restricting purchases of Nvidia chips send someone to lobby Jensen Huang to sell more chips?
He said: "Unpublished means unfalsifiable."
As long as the Justice Department did not produce an indictment, the allegations from above would remain at the level of a war of words, and Nvidia's board would have every reason to protect its founder. To make the board abandon Jensen Huang, what was needed was something that did not depend on intelligence agencies—something that would be an unambiguous violation at any U.S.-listed company.
And so, at 5:31 PM on Wednesday, November 11, the final blow came.
The New York Times published its fourth report. This time the sources were five former Nvidia employees, and the woman was Kimberly Powell, the healthcare business vice president who had been Jensen Huang's direct subordinate for 9 years. The most damning detail in the report: during the GTC conference in San Jose this March, the two were photographed by passersby behaving intimately in a hotel, and Jensen Huang's security team intervened on the spot, demanding that the photographer sign a non-disclosure agreement.
With that, the nature of the matter changed completely.
Entanglement with an external client is at most a personal moral failing; but concealing a years-long conflict of interest with a direct subordinate, and using company security resources to cover it up, is called a breach of fiduciary duty at any U.S.-listed company. McDonald's Steve Easterbrook, Intel's Brian Krzanich, HP's Mark Hurd—all fell on the same red line.
More than twenty minutes later, The Wall Street Journal published its exclusive report that the board was preparing to replace its leader.
Can a major exclusive be written in just over twenty minutes? Perhaps before The New York Times even published, people inside the board—or close to it—had already handed the decision to cut Jensen Huang loose to the media.
Why Did No One Save Him?
Washington hawks handed over the knife, but for things to reach the point where the board was preparing to replace him, another group's acquiescence was indispensable: Jensen Huang's biggest customers, the startups he invested in, and his peers in Silicon Valley.
For the past three years, no one in Silicon Valley dared offend Jensen Huang. That was the era when Larry Ellison and Musk, at Nobu in Palo Alto, spent entire meals pleading with him: "Please take our money, just give us more GPUs." Cards were not evenly distributed—whoever got them first could train their model three months earlier, and the power of life and death was entirely in Jensen Huang's hands.
But the Silicon Valley giants who had long suffered under Jensen Huang had already begun quietly looking for ways out.
One way out was in Washington. According to The Wall Street Journal, Microsoft and Amazon AWS had been lobbying Capitol Hill for the GAIN AI Act since late 2025, with one core demand: AI chip manufacturers must prioritize orders from U.S. domestic buyers before shipping overseas.
The reason Jensen Huang desperately lobbied the White House to relax exports to China—and even supported Southeast Asian cloud vendors like Megaspeed—was not just to sell more cards. The more core business consideration was that he had to retain enough buyers outside the four major U.S. cloud providers, so that pricing power would not fall into the hands of a few giants. For Microsoft and Amazon, supporting export controls under the banner of "national security" had the most direct benefit: legitimately keeping for themselves the production capacity Nvidia had originally allocated for overseas markets.
The other way out lay at the technology foundation.
In the very week Nvidia's market cap evaporated $400 billion in two days, Anthropic—a company Nvidia had once invested in—launched the largest IPO roadshow in history, with a valuation of about $1.4 trillion. Its CEO, Dario Amodei, had always been the most vocal opponent in Silicon Valley of Nvidia's exports to China, having compared selling H200s at Davos to "selling nuclear weapons to North Korea," a characterization Jensen Huang called "stupid."
Anthropic co-founder and chief compute officer Tom Brown revealed that the company used its own frontier model, Claude Mythos, to automatically rewrite and port nearly 100% of its underlying compute kernels in just about 3 days, getting Claude running on AMD chips with almost no engineer intervention—work that previously would have taken dozens of systems engineers more than half a year.
For the past two decades, Nvidia's gross margin of up to 75% rested not on chip hardware as its true moat, but on the extremely high migration costs created by the CUDA ecosystem. But now, AI itself had torn down that wall. Anthropic said in its prospectus that it already treats Amazon Trainium, Google TPU, Nvidia GPUs, and AMD GPUs as a single pooled compute resource. Meanwhile, in-house chips from Google, Amazon, and Meta all scaled up across the board this year.
If this had been 2023 or 2024, shaking Jensen Huang could have crashed the entire U.S. stock market. But by November 2026, Blackwell and Rubin schedules were booked out 18 months; Big Tech's in-house chips were in mass production; even CUDA's migration costs were being flattened by large models.
The capital market's vote was more honest than any statement. On Monday, November 9, Nvidia fell nearly 9% intraday, but the same day the Philadelphia Semiconductor Index rose nearly 7%, AMD surged 13.26%, and Broadcom, TSMC, and SK Hynix all closed higher.
The money had not left AI. The money just felt it no longer necessarily had to bet on Nvidia.
As for Musk, after the November 5 White House Situation Room meeting, before approving a crackdown on chip transshipment companies, Trump had specifically called Musk for his opinion, and Musk expressed support.
Six days later, when the finance account ZeroHedge on X jokingly suggested that Musk hire Jensen Huang, Musk replied an hour and a half later: "That would be an honor."
Wife Guy and 6 Votes
Even so, this still seemed very strange by Silicon Valley standards.
After all, Silicon Valley has always shown astonishing tolerance for the personal morality of top founders. Musk had twins with a female executive at his company Neuralink and was reported to have affairs with multiple subordinates, yet not only did he lose no position, but Tesla's shareholder meeting passed his compensation plan worth up to $1 trillion last year with more than 75% approval. Paul Graham wrote "Founder Mode," and Marc Andreessen railed in his manifesto against "tech ethics" and "risk management"—both were essentially endorsing one thing: as long as the founder can lead everyone to win, the rules can give way for him.
By the standards of Founder Mode, Jensen Huang—with no COO, having 60 executives report directly to him, and working even on Thanksgiving—should have been Silicon Valley's most revered specimen. When the first affair broke, Sequoia partner Shaun Maguire and Bill Ackman did indeed speak up for him on X, saying he was "irreplaceable for American AI leadership."
So why did this exemption suddenly fail by the second affair?
One reason was the public image Jensen Huang himself had cultivated for thirty years.
Musk could smoke weed on a podcast, Andreessen could publicly attack moral censorship, but a Tainan-born Chinese immigrant who was sent to a Kentucky boarding school at age 9, scrubbed men's restrooms, and waited tables at Denny's had no capital for任性—no room for willfulness—if he wanted to be accepted by the power centers of Silicon Valley and Washington.
For thirty years, no matter how high the market cap climbed, Jensen Huang always told the same nearly flawless story to the outside world: he met his first love and wife Lori at 17 and promised to become a CEO before 30; his wife and daughter pick his clothes; apart from work, he goes home to be with his wife and children; even when California's rich collectively protested the wealth tax, he would obediently state that "paying $8 billion in taxes is my honor."
As the American magazine Slate wrote in a commentary on November 11, Jensen Huang fell into the "Wife Guy" trap. Zucker
