On November 8, Jensen Huang published a letter to employees on Nvidia’s official website acknowledging that he had engaged in an extramarital affair. He also said that Nvidia’s board had launched an independent investigation and that there was, at present, no reason to believe US national security or company secrets had been compromised.
What can be established as of November 9 is that the materials made public by US authorities have not substantiated several of the questions and allegations raised against Huang.
To understand the controversy, we must first consider the increasingly stringent restrictions the United States has imposed on China’s artificial intelligence sector in recent years. Far from an isolated incident, it is another dispute arising from Washington’s broader effort to control advanced chips, AI computing capacity, and the technological services associated with them.
Beginning under the Biden administration, the United States steadily tightened restrictions on China’s access to advanced-node semiconductors and AI computing power. In 2021, the National Security Commission on Artificial Intelligence issued its final report, stressing that the United States must maintain its technological lead in AI.
In other words, Washington has defined AI development as a competition it cannot afford to lose. There are, however, two ways to win a race: improve your own capabilities and run faster, or hold back your competitors so they cannot catch up. Consider the old video game Road Rash. Players could race their motorcycles normally, but they could also kick and punch rivals off the track. Knock out every opponent, and you remain in first place no matter how slowly you ride.
A central feature of US policy toward Chinese AI has been its reliance on the second approach. Washington seeks not only to preserve its advantage through innovation but also to deny China the critical resources needed for AI development through export controls, investment restrictions, and technological blockades.
In April 2021, the US Commerce Department’s Bureau of Industry and Security (BIS) added seven entities—including Tianjin Phytium, Sunway, and several Chinese national supercomputing centers—to its Entity List, citing their involvement in China’s military modernization. That June, the Biden administration issued Executive Order 14032, further revising restrictions on securities investments in certain Chinese defense-related companies.
In 2022, the United States enacted the CHIPS and Science Act to support domestic semiconductor manufacturing and research. At the same time, its subsidy guardrails restricted recipients from expanding advanced semiconductor capacity in China and other so-called “foreign countries of concern,” while imposing constraints on sensitive technology research, licensing, and transfers.
On October 7, 2022, the Commerce Department introduced further export controls covering advanced computing chips, supercomputer-related end uses, advanced semiconductor manufacturing equipment, and certain activities by US persons supporting the production of advanced chips in China.
These measures affected Nvidia’s A100 and H100 chips and related systems. Washington also invoked the Foreign Direct Product Rule to bring certain goods manufactured abroad using US technology or software within the scope of American export controls.
The objective, then, was not merely to prevent China from purchasing advanced computing hardware directly. It was also to impede China’s ability to expand domestic production of sophisticated chips. From procurement and manufacturing equipment to technical assistance, Washington was assembling an increasingly elaborate system of restrictions.
In 2023, the United States tightened the performance thresholds governing advanced computing chips, bringing even Nvidia’s downgraded A800 and H800 products, designed specifically for the Chinese market, under the restrictions. It also stepped up scrutiny of overseas end users and pressed countries including Japan and the Netherlands to cooperate on export restrictions for semiconductor manufacturing equipment.
In 2024, the controls expanded to high-bandwidth memory, semiconductor manufacturing equipment, and software tools. On December 2, the Commerce Department announced new restrictions covering 24 categories of semiconductor manufacturing equipment, three categories of software tools, and high-bandwidth memory, while adding 140 entries to the Entity List.
Meanwhile, US oversight increasingly extended to investment activity and overseas computing services. Restrictions on certain outbound US investments involving semiconductors, quantum information technologies, and AI in mainland China, Hong Kong, and Macao took effect in January 2025.
Although ostensibly designed to mitigate risk, these measures reveal a broader trend: Washington is stretching the concept of national security ever further, drawing ordinary economic activity and technological competition into its security-control apparatus.
When Donald Trump returned to the White House in January 2025, the policy underwent some adjustments. Whereas the Biden administration had sought to establish a global control framework, the Trump administration placed greater emphasis on country-specific arrangements, case-by-case licensing, and the balance between restrictions and commercial interests. None of this meant that Washington had abandoned its efforts to contain China technologically.
Before leaving office, the Biden administration introduced an AI Diffusion Framework that sought to apply different licensing requirements to advanced chips according to their destinations and to regulate the model weights of certain leading closed-source AI systems. Among its principal objectives were preventing advanced chips from reaching China through third countries and stopping Chinese firms from obtaining the corresponding computing services through overseas data centers.
In May 2025, the Trump administration announced that it would not enforce the framework’s new requirements and initiated the process of rescinding them. Yet in May 2026, the US Government Accountability Office pointed out that the Commerce Department had not formally completed the rescission; instead, it had adopted a general policy of non-enforcement.
That left a regulatory gray area in which the US government could revive, strengthen, or modify the restrictions whenever political circumstances demanded.
Nvidia’s H20 chip offers a particularly revealing example of this policy volatility. In April 2025, Washington required Nvidia to obtain licenses before exporting H20 chips to China. Nvidia subsequently recorded approximately $4.5 billion in charges related to excess inventory and purchase commitments. By August, US authorities had begun issuing licenses for transactions involving some Chinese customers. Nvidia, however, said that Chinese buyers had little interest in imported chips whose performance fell short of their needs, limiting sales of the remaining inventory.
On January 15, 2026, the United States changed its licensing review policy for certain exports of H200-class chips to China from a presumption of denial to case-by-case review. Qualifying products, including Nvidia’s H200 and AMD’s MI325X, thus became theoretically eligible for export-license applications.
But case-by-case review is hardly the same as unrestricted exports. Applicants must demonstrate that domestic US supply is sufficient and that the proposed exports will not divert production capacity away from the American market. Buyers must have appropriate screening and security procedures, and the products must undergo testing by an independent third party in the United States. Washington also requires disclosure of certain remote end users and imposes aggregate performance caps on relevant chips shipped to mainland China or Macao. More powerful products remain outside the scope of the relaxation.
There is therefore a considerable gap between being permitted to apply for a license and actually delivering a product. US policy may appear to have softened, but its underlying logic remains fundamentally unchanged. At the same time, the growth of China’s AI industry is reshaping the market. As Chinese technological and industrial capabilities improve, both Chinese companies’ dependence on downgraded American chips and their negotiating strategies are changing.
US efforts to control the AI supply chain are also moving beyond chips themselves to encompass AI models and cloud computing services. Whereas regulators once focused primarily on whether Chinese companies could purchase advanced chips directly, they are now paying closer attention to compliance issues involving data-center services in Southeast Asia.
This helps explain why data-center operations in countries such as Singapore and Malaysia have increasingly attracted US regulatory scrutiny. As early as 2025, individuals in Singapore faced charges in a case involving alleged misrepresentations about the end users of servers. Similar compliance controversies have arisen in Malaysia.
The first essential backdrop to the Jensen Huang controversy, then, is Washington’s long-running campaign to tighten its control over China’s access to advanced chips, computing services, and other technologies it regards as strategically important. Such controls have become a major component of US national security policy.
More revealing, however, is a second backdrop: the convergence of America’s ever-expanding national security rationale, the broader strategic rivalry between China and the United States, and a McCarthyite political atmosphere at home.
All three congressional committees that demanded a closed-door briefing from Jensen Huang are chaired by Republicans. Yet on China-related issues, Democrats and Republicans alike can adopt highly securitized positions. Anxiety about China in American domestic politics has reached an almost absurd level. A person’s Chinese ancestry, a company’s business dealings with China, or even routine commercial interactions with the Chinese government may be enough for some politicians and media outlets to insinuate an intelligence risk.
This line of reasoning is hardly new. Some American politicians have long relied on patently absurd arguments to allege links between Chinese companies and the Chinese military. Now similar presumptions are being directed at Huang: the Chinese heritage of those involved, their business connections with China, and their private relationships have been woven by the media into a narrative saturated with national security implications.
The problem is obvious. Commercial ties do not, by themselves, constitute evidence of unlawful conduct. Still less can a person be branded an agent of a foreign intelligence service on the basis of ethnicity or unsubstantiated associations.
In this sense, Jensen Huang is facing a classic national security witch hunt in the United States. Beneath it lies a deeper problem: Washington’s mounting strategic anxiety over the development of China’s AI industry. The United States initially assumed that controlling advanced-node chips and restricting China’s access to high-end computing power would effectively stifle Chinese AI development. Events, however, have not unfolded entirely as Washington expected.
Despite continuing US technological restrictions, Chinese open-source large AI models have made repeated breakthroughs, delivering impressive performance at lower cost and with greater efficiency. They have also gained increasing recognition in real-world applications and market adoption.
The practical challenges to Washington’s strategy of securing a lasting lead through technological blockades are therefore becoming harder to ignore. The United States has cast AI competition as a race it must win, but China’s progress demonstrates that holding back a rival does not necessarily guarantee victory.
As Chinese AI companies continue to achieve breakthroughs, some political forces in the United States are reluctant to acknowledge the powerful internal drivers of Chinese technological innovation and industrial growth. Instead, they seek an alternative explanation: Were these advances made possible by American chips obtained in violation of export controls, or by computing services routed through overseas data centers?
They then seek to elevate what would ordinarily be export-control compliance questions arising from commercial activity into matters of national security or even espionage. By invoking the supposed threat from China, Washington can justify ever tighter technological restrictions while, to some extent, obscuring the failure of earlier policies to deliver their intended results.
If transshipment or smuggling in violation of export controls has indeed occurred, the authorities can investigate on the basis of evidence and pursue the matter under the law. But why must the compliance risks inherent in ordinary commercial transactions be linked to alleged national security threats and foreign intelligence agencies? Once an issue is labeled a national security matter, the agencies involved may gain broader investigative powers and greater room for political maneuver. Alleged wrongdoing that should require specific evidence can more easily become a political controversy in which the accused is continually forced to prove their innocence.
This mindset was already on display during congressional questioning of TikTok CEO Shou Zi Chew. Although Chew is Singaporean, American lawmakers repeatedly pressed him about his ties to the Communist Party of China. The exchanges reflected a deeply entrenched assumption: anyone of Chinese ethnicity may be presumed to have some special connection to the Chinese government.
Applied to Huang, that same logic would expose his personal identity, private relationships, and corporate activities alike to political scrutiny.
There is an even subtler question here. If Jeff Bezos or Bill Gates were embroiled in a comparable controversy over a private relationship, would American politicians and media outlets be equally quick to turn their personal lives into matters of national security?
Could Huang’s particular predicament also reflect the ethnic prejudice that has long existed in American society? The intersection of America’s McCarthyite anti-communist political tradition, its deeply securitized view of China, and questions of ethnic identity may expose Huang to an especially complex set of political risks.
Of course, how the affair develops will depend on the investigation’s findings and the evidence that emerges. But if American domestic politics continues in this direction, we may witness an uproar not yet supported by sufficient facts gradually being transformed into political pressure over the management—and potentially even the ownership and control of assets—of a major technology company.
If media-driven pressure alone can propel inadequately substantiated allegations against an entrepreneur as influential as Huang, and then shape corporate governance, what becomes of America’s professed commitment to free markets and the protection of private property? This could prove to be a significant turning point.
The Huang controversy exposes more than a clash between domestic politics and technology capital in the United States. It also reveals the anxiety and imbalance of a hegemonic power in relative decline as it confronts external competition. When a country loses confidence in its ability to stay ahead through fair competition and sustained innovation, it may increasingly resort to administrative restrictions, political pressure, and national security reviews to hold its competitors back.
Yet such tactics may do little to resolve its underlying problems. Instead, they risk further disrupting markets, harming businesses, and undermining the very institutional principles the country claims to uphold. In that sense, Huang’s ordeal may once again lay bare the realities of American free-market capitalism when it comes under political pressure.
How will a hegemon that has lost its bearings—and an America increasingly unsure of its ability to win through ordinary competition—respond to the rise of China’s AI industry? History may already be providing an answer. Washington’s attempt to preserve its advantage by restraining its competitors may ultimately become yet another illustration of its own relative decline and policy failure.
Editor: Chang Zhangjin
