Jensen Huang's troubles are just beginning

Nvidia had originally planned to hold its AI Day Seoul 2026 event on November 9, and Jensen Huang's eldest daughter, Madison Huang, a senior director at Nvidia, was scheduled to attend.

Under the previous arrangement, Madison Huang was to attend the event in Seoul and meet with South Korean government officials to discuss cooperation in artificial intelligence.

South Korea is itself a key semiconductor partner market for Nvidia globally, with companies such as Samsung and SK Hynix maintaining close business ties with Nvidia. Madison Huang has also participated in South Korea's AI industry events on multiple occasions, so this event was in fact fairly important to her.

Then, the day before departure, her entire schedule was suddenly canceled. At the time, Yonhap News Agency, citing sources, said the cancellation might be related to family matters but did not disclose the specific reason.

But looking at it now, the reasons for Madison Huang's cancellation involved both personal and professional matters.

The personal matter was that her 63-year-old father had an affair with a Chinese woman named Huang Le, who was once an executive at the Singapore-based data center company Megaspeed.

On November 8, Jensen Huang personally sent an apology letter to all Nvidia employees. In the letter, the founder of the tech giant, who has led Nvidia for more than thirty years, publicly admitted to being unfaithful to his wife.

He said his behavior had hurt his wife and children, that there was no excuse for it, and he apologized to his family, employees, the board, and shareholders.

In fact, Jensen Huang met his wife Lori in college, and the two have lived together for decades and have a son and a daughter. Both children later went on to work at Nvidia. Now, this tech entrepreneur once seen by the outside world as a success in both career and family has been forced to publicly confront a crisis in his marriage.

As for the professional matter, it is most likely the trouble facing Nvidia.

On November 7, the Financial Times, citing people familiar with the matter, reported that an internal memorandum previously obtained by U.S. intelligence showed that Huang Le, the woman romantically involved with Jensen Huang, was suspected by U.S. intelligence agencies of having been recruited by Chinese intelligence to approach Jensen Huang and obtain information related to Nvidia and advanced U.S. AI chips.

Nvidia's board was in fact also overwhelmed. It formed a committee of independent directors and hired outside counsel to investigate whether Jensen Huang had been targeted by a foreign intelligence operation.

As a senior director, Madison Huang naturally also had to return to help deal with this mess.

At this point, Jensen Huang's affair had already far exceeded the scope of his personal family life.

Behind this matter lies in fact a very complex logic of maneuvering and rivalry, and Jensen Huang's troubles may only be just beginning.

What exactly is the story behind Megaspeed?

Why did U.S. intelligence agencies target Huang Le? This matter actually has to start with Megaspeed.

Megaspeed initially had nothing to do with artificial intelligence.

In 2015, a company named 7Road International was incorporated in Singapore, backed by the Chinese game company 7Road. 7Road was once an important player in China's browser game industry, having developed games such as DDTank and Wartune, and later received support from investment institutions with state-owned capital backing.

In 2023, as ChatGPT set off a global artificial intelligence boom, the Singapore company's business direction changed.

In March of that year, 7Road International officially changed its name to Megaspeed International and began entering the AI computing power and cloud computing industry. Huang Le, mentioned earlier, also became the company's managing director during this period.

After that, the company's equity was transferred to the Singaporean company Swiftdata, and Huang Le exited the list of directors. In early 2024, Megaspeed set up a subsidiary in Malaysia called Speedmatrix, responsible for local equipment procurement and related business.

This company, which had previously had almost no reputation in the global tech industry, then took on a business of astonishing scale.

In June 2024, Speedmatrix began large-scale procurement of AI servers equipped with Nvidia GPUs.

According to a New York Times investigation of ImportGenius trade records, between June 2024 and February 2026, Megaspeed's Malaysian affiliate purchased at least $4.8 billion worth of advanced AI equipment, including large quantities of Nvidia chips and related technology.

For a company that had only recently pivoted to AI, this scale of procurement was quite staggering.

And its procurement channels also drew the attention of U.S. regulators.

A considerable portion of the equipment Megaspeed purchased came from the U.S. server supplier Aivres, which in turn has close ties to the Chinese server manufacturer Inspur Group.

Some entities under Inspur Group were added to the U.S. Entity List in 2023 over support for China's military supercomputing development.

But Aivres, as a company operating in the United States and not placed on the same restricted list, was still able to purchase Nvidia chips while complying with relevant export rules.

Thus, a cross-border supply chain took shape.

Nvidia supplies the GPUs, server suppliers such as Aivres handle equipment integration, and the related products are then exported to Speedmatrix in Malaysia.

These servers were subsequently used for data center business in Southeast Asia.

Megaspeed's business model mainly involves renting out GPU computing power to other companies. It buys chips and servers, deploys the equipment in data centers, and then provides remote computing services to customers.

Customers do not need to buy chips; they simply access computing resources over the network.

Megaspeed's data centers in Malaysia and Indonesia have provided computing services to Chinese companies, including Alibaba.

This touches on an issue in U.S. chip export controls.

The United States began restricting exports of certain advanced AI chips to China in 2022, but these restrictions do not mean that all Chinese companies are barred from using computing resources located overseas.

If a Chinese company rents GPUs through a data center in Malaysia and the chips remain in Malaysia the whole time, whether the transaction violates U.S. rules must be judged in light of export licenses, end users, and the actual transaction arrangements.

But if a company reroutes controlled chips that should have been deployed in Malaysia to China without authorization, that could directly violate U.S. export control regulations.

U.S. investigators suspect that some of Megaspeed's business may involve the latter scenario.

In fact, the U.S. government's attention to Megaspeed began well before the revelations about Jensen Huang's personal relationship.

In 2024, officials from the U.S. Bureau of Industry and Security traveled to Malaysia to inspect a data center used by Megaspeed. Investigators found that some of the purchased Nvidia equipment was still in its packaging and had not been put into operation, which aroused suspicion.

By October 2025, the New York Times published an investigative report on Megaspeed.

That same month, Republican Senator Tom Cotton formally wrote to U.S. Commerce Secretary Howard Lutnick, demanding an investigation into whether Megaspeed used complex corporate relationships and cross-border transactions to help Chinese customers circumvent U.S. restrictions on advanced chip exports.

Cotton specifically mentioned the company's ties to suppliers linked to Inspur Group and the difficulty of confirming the final use of the chips.

As the investigation deepened this year, the U.S. Commerce Department's probe into Megaspeed was transferred to the Department of Justice, and the matter entered the scope of U.S. judicial investigation.

Faced with U.S. questions, Nvidia has consistently maintained its position.

On October 29, Jensen Huang said in an interview with Bloomberg that Nvidia's business dealings with Megaspeed were entirely normal. He also stressed that the company had checked Megaspeed's GPU deployment, verified the relevant equipment, and found no evidence that the chips had been illegally transshipped to China.

So based on the information publicly available so far, U.S. investigators are mainly focused on two things:

First, whether the advanced AI equipment Megaspeed purchased was illegally transshipped to China.

Second, whether the computing services provided by this equipment involved helping restricted entities evade U.S. export rules.

The investigation has not yet reached a final conclusion, but Megaspeed's experience already reflects the real problems facing U.S. chip controls.

Even if chips are restricted from being exported to a certain country, local companies may still be able to obtain computing power through data centers in other countries.

And the reason the United States places such importance on these chips goes back to the chip blockade against China that began several years ago.

The U.S. chip blockade against China

On October 7, 2022, the U.S. Commerce Department announced a new set of export control measures.

Under the rules, exports to China of advanced computing chips such as Nvidia's A100 and H100 required prior U.S. government authorization. Certain semiconductor manufacturing equipment and supercomputing-related technologies were also brought within the scope of the restrictions.

Previously, the United States had already used tools such as the Entity List to restrict Chinese tech companies like Huawei from obtaining certain advanced technologies. But this time, the restrictions were expanded to advanced computing chips and related industrial chains, directly affecting the Chinese AI industry's access to high-performance computing power.

The U.S. Commerce Department cited national security as the reason.

The U.S. government believes that advanced computing power can not only train AI models but also be used for military simulation, intelligence analysis, weapons development, and supercomputing. If China can continue to obtain the most advanced U.S. chips, it could narrow the gap between the two countries in related technology fields.

Therefore, the United States hopes to slow China's development in these fields by restricting the supply of advanced chips and semiconductor manufacturing technology.

But this policy soon ran into Nvidia.

At the time, Nvidia already occupied an important position in the global AI computing market, and China was one of its important markets.

To maintain its China business, Nvidia began adjusting product designs and launched chips such as the A800 and H800 that complied with the export rules in place at the time.

These products had limits placed on certain performance metrics, allowing them to meet U.S. export requirements at the time while continuing to be sold to Chinese customers.

However, the U.S. government soon discovered that Chinese companies could still use these products to build large-scale AI computing clusters.

On October 17, 2023, the Biden administration further tightened controls, adjusting the performance thresholds and scope of application for advanced chips, and the A800 and H800 that Nvidia had specifically launched for the Chinese market were also subject to new restrictions.

Nvidia subsequently launched products such as the H20, continuing its attempts to preserve the Chinese market.

This created a situation that lasted for years: the U.S. government kept adjusting export rules, and Nvidia kept modifying product designs in response, trying as much as possible to keep its China business running.

By 2025, Trump returned to power, and Nvidia had initially hoped the U.S. government would ease some restrictions, but things did not go smoothly.

On April 9, the U.S. government notified Nvidia that exports of H20 chips to China also required a license, directly affecting Nvidia's already arranged production plans and customer orders.

According to a filing Nvidia submitted to the U.S. Securities and Exchange Commission, the company recognized about $4.5 billion in charges in its first fiscal quarter of that year due to H20 inventory buildup and purchase commitments.

A month later, at Computex Taipei, Jensen Huang publicly expressed dissatisfaction with U.S. export policy.

He argued that continuing to restrict U.S. companies from selling AI chips to China was pushing Chinese customers toward domestic suppliers such as Huawei.

Jensen Huang revealed at the time that Nvidia's share of China's AI chip market had fallen to 50% from about 95% previously.

He also predicted that China's AI market could reach $50 billion in 2026. In his view, if U.S. companies withdrew from this market over the long term, what they would lose was not only chip sales revenue but also the future dependence of China's AI industry on the U.S. technology ecosystem.

This also touches on Nvidia's own business model.

One of Nvidia's most important competitive advantages is the software ecosystem it has built over the years: CUDA.

Developers using Nvidia GPUs to train and deploy AI models often need the accompanying software, development tools, and technical support.

As more and more companies use Nvidia products, the related software and computing infrastructure also continue to develop around them, and this ecosystem helps Nvidia maintain its long-term competitive edge.

China, meanwhile, has a vast population of internet companies, AI developers, and researchers.

If these users cannot buy Nvidia chips for a long time, they will have a stronger incentive to look for alternatives.

Once domestic AI companies gradually adapt to local computing platforms such as Huawei's, even if the United States later reopens some chip exports, Nvidia may not easily win back its former market.