Jensen Huang is out. What did China want from him?

The Verge's Decoder spoke with Kyle Chan of the Brookings Institution's John L. Thornton China Center just before Nvidia's CEO resigned. The conversation asks the question most coverage skips: what would China actually get from a source so close to the head of Nvidia?

Chan's answer is not about stolen designs. A CEO who travels the world giving policy speeches is not the obvious target for chip specs. The likelier prizes are insight into US policymaking and, in the most worrying case, influence over export controls, through someone the president reportedly calls directly. That moves the story from personal scandal to the integrity of national technology policy.

We appreciate how carefully Chan separates allegation from fact. US intelligence reportedly believes Alice Huang may be an asset of Chinese intelligence. Megaspeed is an Nvidia customer that built a data center in Malaysia and may be a conduit for Chinese access to advanced chips like Blackwell. Those are serious claims, and they are still claims. We should hold that line too.

What strikes us is how little the supposed plot adds up on its own terms. Nvidia was already lobbying to loosen export limits for its own commercial reasons. Chips were allegedly already flowing through intermediaries. So Chan floats a less cinematic explanation: Beijing may simply have wanted visibility, a window into a confusing situation, rather than a grand strategy. We find that persuasive, and it is a useful corrective to the idea of a perfectly coordinated Chinese state. Chan describes a system full of turf wars, corruption, and missteps that can backfire badly.

And it may be backfiring. The episode has destabilized the American AI economy, which looks like a win for China, yet it is also likely to harden Washington's stance. Pending export control bills could gain momentum. The industry's most powerful advocate for diffusion over restriction is gone. If China's goal was softer US policy, this could deliver the opposite.

The part we cannot get past is the US side. Reporting cited in the episode says intelligence agencies flagged concerns roughly a year ago, a memo was overseen by the secretary of state, and nothing happened. Nvidia's spokespeople called the relationship normal on the record shortly before the CEO's apology letter acknowledged it. That pattern, dismissal then reversal, is what erodes trust in both a company and a government. A shareholder suit has already been filed.

Our take: China's AI strategy is less about the frontier and more about spreading AI through its economy, especially manufacturing, with domestic chips from firms like Huawei that are good enough. Washington's answer cannot be built on personal access to one executive. It has to rest on transparent, enforced rules.

Chan argues China has spent years pushing its AI companies off Nvidia and toward domestic chipmakers, aiming for a supply chain that is resilient to US policy even if it never fully catches up. Many Chinese AI firms had been major contributors to Nvidia's CUDA software ecosystem before that shift.