Jensen Huang resigns and the AI empire wobbles

Jensen Huang resigned as chief executive of Nvidia on Thursday, ending a thirty-year reign at the company he co-founded and transformed into the most strategically consequential technology corporation on earth. The BBC reported that the resignation followed weeks of mounting pressure from the board after the FBI confirmed it was investigating an alleged romantic relationship between Huang and a woman US intelligence believes was an operative of Chinese state intelligence services. Al Jazeera confirmed that Colette Kress, Nvidia’s long-serving chief financial officer, has been appointed interim chief executive while the board conducts a formal succession process. The company’s shares fell sharply at the open before recovering partially. NPR’s analysis of the so-called “honey trap” tactic used against Huang describes it as a textbook intelligence operation: cultivating a relationship over time with a high-value target, harvesting information incrementally, and relying on the target’s combination of ego and discretion to prevent disclosure.

The received wisdom

The consensus framing in the financial and technology press is sympathetic to Huang’s legacy while treating the resignation as a regrettable personal failing that the company will ultimately absorb. Nvidia’s products remain irreplaceable, the argument goes; the H100 and its successors have no serious Western rival; the structural demand for AI computing capacity will not diminish because of one man’s poor personal judgment. The corporate governance school adds that the board acted correctly and swiftly, that the internal investigation process demonstrates institutional resilience, and that the transition to Kress — a respected operational executive — provides continuity. The framing is essentially one of a great institution surviving a scandal, as great institutions have survived scandals before. IBM survived its antitrust decades. Intel survived its architecture stagnation. Nvidia will survive Jensen Huang’s affair. The technology moves on; the talent disperses; the stock recovers.

A different read

This framing is too comfortable by half, and it papers over several things that deserve harder examination.

Start with the intelligence dimension. The woman at the centre of the FBI’s investigation — identified in multiple reports as Alice Huang, no relation — was not a chance encounter. The Guardian’s Thursday briefing notes that intelligence professionals describe her as a “long-cultivated asset,” someone who had been in Huang’s social circle for years before any romantic relationship developed. This is the patience of the operation that is most alarming. Chinese intelligence does not improvise honey traps; it engineers relationships over years, embedding agents in the social ecosystems of high-value targets long before any operational objective is activated. The question is not just what Huang may have disclosed — it is how many other executives in the semiconductor and AI supply chain are similarly situated, and how many of those relationships have not yet come to light.

Second, consider what Nvidia actually is. It is not merely a technology company. It is a critical national infrastructure node in a way that no previous civilian corporation has quite been. The H100 chip is the operating substrate of the global artificial intelligence race. Governments from Washington to Riyadh to New Delhi have structured their AI strategies around access to Nvidia’s product. Export controls on Nvidia chips to China have been among the most consequential technology policy decisions of the past five years. The idea that the chief executive of such a company was — potentially — under the influence of an intelligence operative for an extended period is not a corporate governance story. It is a counterintelligence crisis of the first order.

Third, there is a pattern here that the tech industry keeps refusing to confront honestly. The BBC’s analysis of Huang’s rise and fall notes that Silicon Valley’s culture of founder-worship creates structural vulnerabilities: boards defer to visionary CEOs long past the point when independent oversight would check their excesses. In Huang’s case, the personal protection around him — the retinue, the adulation, the assumption that normal rules did not apply — appears to have made routine counterintelligence briefings and relationship disclosure requirements feel beneath him. This is a pattern with historical precedents. Robert Hanssen, the FBI agent who spied for the Soviets for over two decades, was enabled partly by an institutional culture that made him feel too valuable to scrutinise. The mechanism differs, but the dynamic of institutional deference enabling personal compromise is recognisable.

The deeper problem is structural. America has made its most sensitive technology dependent on a handful of civilian companies led by individuals who are not subject to the security clearance requirements, the lifestyle polygraphs, or the counterintelligence protocols that would govern a government official with equivalent access to national secrets. The Huang affair — like the earlier questions about relationships between Chinese venture capital and American AI startups — is a symptom of a larger failure to apply Cold War-era security logic to the new economy where the decisive competition is happening.

What to watch

Watch whether Congress moves to extend security clearance requirements or mandatory counterintelligence briefings to the C-suite executives of companies that handle export-controlled technology. Watch whether the Nvidia board chooses a permanent successor who is primarily a technologist or primarily an operator with a strong legal and compliance background — the choice will signal whether the board has absorbed the lesson. Watch the share price trajectory: if it recovers fully within sixty days, the market will have decided that governance risk at AI infrastructure companies is not a systemic concern, which is itself a systemic concern. And watch whether the FBI investigation produces criminal charges, or whether it quietly closes — because the political incentives to avoid a full accounting of what was actually disclosed may prove stronger than the counterintelligence incentives to surface it.

— J