On 12 November, Nvidia’s board announced that Jensen Huang had resigned as president, chief executive and board member, effective immediately, ending 33 years at the head of the company he co-founded. Chief financial officer Colette Kress was named interim CEO, and Harvey Jones became the board’s first-ever chair.

Most of the coverage has focused on the personal side of the story and on what it means for Nvidia’s share price. This piece looks at something else: what the documented record shows about how US export controls on advanced AI chips are enforced in practice, where that system depends on the companies it regulates, and which questions the public record cannot yet answer. Those issues stand regardless of how the investigations into the individuals involved conclude.

What happened

On 8 November, Huang acknowledged an affair in an email to Nvidia staff. The same day, the board opened an independent investigation into whether he had been the target of a foreign intelligence effort. The investigation followed reporting by The New York Times and the Financial Times on his relationship with Alice Huang, a former executive at Megaspeed International, a Singapore-based Nvidia customer. The two are not related.

At that point, an Nvidia spokesperson said Huang was committed to continuing as chief executive. Over the following days, further reports about his personal conduct emerged, and on 12 November the board announced his resignation. Nvidia’s shares closed down 4.34 per cent that day, having already fallen more than 20 per cent over the previous month.

The allegations and the denials

According to CNN, a classified US intelligence memo issued in May concluded that Alice Huang was a Chinese intelligence asset. NBC News separately reported, citing a person familiar with the matter, that US intelligence agencies had reached the same conclusion. The New York Times has reported that Jensen Huang was the target of a Chinese intelligence operation that sought to send advanced Nvidia AI chips to China.

These claims have been denied or remain unconfirmed. China’s foreign ministry has called the allegation groundless and a smear. Megaspeed has categorically denied diverting chips. Before the resignation, Nvidia’s board said it had no reason to believe company secrets or national security had been compromised. President Trump has said federal investigators are examining the national security implications. No charges have been reported against any individual, and the intelligence assessment has not been made public.

The espionage allegation, in other words, currently rests on unnamed officials and a classified document. The rest of this piece does not depend on it.

Who is Megaspeed?

Megaspeed International describes itself as a Singapore-based cloud provider and an Nvidia Cloud Partner in the Asia-Pacific, with operations in Singapore and Indonesia. According to company filings reported by Briefs, it held about $3 billion in assets in 2024, on revenue of $294 million and a loss of around $20 million. The same report notes that the company previously named data centres in Malaysia and Indonesia before relabelling their locations simply as “East South Asia.”

The firm reportedly became Nvidia’s largest chip buyer in Southeast Asia, purchasing more than $4.8 billion worth of its chips over two years, according to New York Times reporting cited by Fortune. Its business model matters for what follows: as CNBC has explained, a cloud provider of this kind can give customers remote access to computing power from advanced chips located outside China.

What the Megaspeed record shows

The US government sets export controls on advanced AI chips, but much of the day-to-day checking of where chips end up falls to the companies that sell them. The Megaspeed case shows how that works in practice.

Nvidia’s public position has been consistent. In October 2025, a spokesperson said the company had visited Megaspeed’s operations and found no evidence of diversion. Last month, Nvidia described the diversion allegations as having been thoroughly investigated and debunked. Megaspeed, for its part, says inspections by US, Malaysian and Nvidia teams found no violations.

Other reporting points in a different direction. Nvidia told the US government that its inspections had found Megaspeed’s hardware where it was supposed to be, but trade records analysed by Bloomberg indicated that Nvidia’s calculations had undercounted chips sold to Megaspeed by around 50,000 units. Bloomberg also reported that Nvidia’s general counsel intervened in the US government’s investigation into the company. Megaspeed nonetheless remains an official Nvidia Cloud Partner. Separately, Singapore police opened an investigation in October 2025 into whether Megaspeed had broken domestic law; no findings have been published.

Taken together, these facts show a specific structural feature: the company whose sales were under scrutiny was also one of the main sources of information about whether those sales complied with the rules. That does not establish wrongdoing by Nvidia. It does show that the oversight arrangement rested heavily on the seller’s own checks, and that independent trade data has now called those checks into question.

Not an isolated case

Megaspeed is not the only company to face scrutiny over where Nvidia hardware ends up. Fortune reports that a separate case involves $2.5 billion in servers equipped with Nvidia chips connected to Super Micro, one of Nvidia’s closest partners. Singapore has also previously investigated suspected fraud in server shipments, with its law minister saying authorities had assessed that the servers may contain Nvidia chips subject to US export restrictions.

These cases differ in their details, and none has produced a published finding that Nvidia itself broke the law. What they share is the route: high-value hardware sold to intermediaries and partners in Southeast Asia, where determining the final user depends on records and inspections that are difficult for regulators to verify independently. Singapore’s weight in that route is considerable. In early 2025, it accounted for about 22 per cent of Nvidia’s reported revenue, although Nvidia has said that revenue associated with Singapore does not indicate diversion to China.

The remote-access question

The case also draws attention to a gap in the rules themselves. According to New York Times reporting, nearly $2 billion of the chips Megaspeed bought were installed in data centres in Malaysia and Indonesia that are alleged to serve Chinese customers remotely. US export controls restrict the shipment of certain chips to China, but remote access to computing power located elsewhere is treated as a separate question.

This is a feature of the current rules rather than a finding about Megaspeed. The controls are intended to limit China’s access to advanced computing for AI. If computing power located outside China can be used by Chinese customers over the internet, rules focused on physical shipments do not fully cover that objective.

Two governments, one company

Nvidia has been under pressure from both Washington and Beijing for some time, and the record shows each side treating its chips as a strategic lever. In July 2025, US Senators Elizabeth Warren and Jim Banks warned that a planned China trip by Huang could undermine American export controls. The following month, China’s cyberspace regulator summoned Nvidia over what it called “backdoor safety risks” in its H20 chips.

The policy itself has also shifted. The Trump administration has permitted sales of Nvidia’s H200 chips to approved customers in China while keeping restrictions on the company’s most advanced products. Beijing’s response was not straightforward acceptance: in January, Reuters reported that Chinese authorities had told customs agents the H200 was not permitted to enter China, though it was unclear whether this was a formal ban or a temporary measure.

The result is a regime in which some Nvidia chips may be sold to China and others may not, with both governments adjusting their positions. In that setting, enforcement depends even more on accurate information about who the end customer is, which is precisely the information the Megaspeed case has called into question.

The questions that remain open

Several questions will determine how significant this case proves to be, and none can yet be answered from the public record.

The first is timing. If, as reported, US intelligence reached its conclusion about Alice Huang in May, it is not yet known when, or whether, that assessment was shared with Nvidia. That matters because the company continued to defend Megaspeed publicly in the months that followed. The second is the Bloomberg discrepancy: whether the roughly 50,000 uncounted chips reflect an accounting gap or something more is central to the US investigation. The third is the intelligence assessment itself. Until officials release more information or investigators bring charges, the espionage claim cannot be independently assessed. The fourth is Singapore’s investigation, which has been running for more than a year without published findings.

What to watch

Four processes are now running in parallel, and each could change the picture. Nvidia’s board is continuing its independent investigation into whether Huang was the target of a foreign intelligence effort, while also searching for a permanent chief executive from internal and external candidates. US federal investigators are examining both the national security implications of the case and the earlier questions about whether Megaspeed diverted chips. Singapore’s police investigation into Megaspeed remains open. And any findings on the Bloomberg trade-record discrepancy will indicate whether the issue is one of accounting or of enforcement.

Each of these could produce new facts within weeks. Until they do, the most reliable guide is the documented record rather than the allegations surrounding it.

What can be said now

What the record supports is narrower than the headlines, but still significant. Enforcement of US chip controls in Southeast Asia has depended heavily on information supplied by the companies being regulated. Independent trade data has challenged that information in at least one major case. Current rules do not clearly address remote access to restricted computing power. And both Washington and Beijing are actively using access to these chips as leverage, which raises the stakes of every gap in the system.

Nvidia’s leadership change will dominate market coverage for weeks. The more durable story is about the enforcement system, and that story will be shaped less by the scandal than by what the US and Singaporean investigations eventually find.