Without Jensen Huang, who holds the US-China tech bridge together?

The most consequential effect of Jensen Huang’s departure on the China-US AI race may come through export policy. Nvidia retains strong commercial reasons to serve Chinese customers, but must now make that case in Washington without Huang at the helm. Tighter restrictions could raise Chinese developers’ costs while encouraging them to build on alternative platforms.

On 12 November, Nvidia announced Huang’s resignation as CEO and his departure from its board, with CFO Colette Kress named interim CEO. His exit followed his admission of infidelity and an independent inquiry into reports that he had been targeted by a foreign intelligence operation. China has rejected the allegations of Chinese involvement.

Nvidia’s core technological strengths remain in place after Huang’s departure. Its advantage combines advanced processors with CUDA software libraries and development tools, which help developers turn hardware into usable computing power. Customers will judge its reliability through product delivery and support. The broader competitive consequences depend on how political pressure changes access to that platform.

US policymakers face a trade-off. Restricting access to advanced computing could raise the cost of AI development for China. But allowing Chinese developers to continue using American platforms would help preserve those platforms’ commercial reach. Huang has argued that excluding Nvidia from China leaves opportunities for Chinese competitors. His company’s interests overlap with a strategic argument about preserving American technological influence.

Case weakens for selective access for China

The controversy could weaken that argument politically. Congressional leaders have requested briefings on the allegations and Nvidia’s security protocols. If advocating chip sales becomes associated with compromised judgment, officials may find it harder to defend selective access. Kress and any permanent successor will inherit Nvidia’s incentive to retain Chinese customers, but will not automatically inherit Huang’s personal influence in Washington.

For Chinese AI developers, the immediate risk is less predictable access to large amounts of computing power. That affects how many experiments a laboratory can run and how quickly it can train or deploy demanding models. Greater efficiency can help stretch available resources, but repeatedly redesigning plans around uncertain supply consumes money and engineering time.

Competition also extends beyond chip shipments. Nvidia provides tools to deploy and optimise Chinese open-weight models, including DeepSeek and Qwen. A Chinese model can attract users worldwide while generating demand for American computing infrastructure. Progress by Chinese model developers can therefore reinforce part of the US technology ecosystem, even as it challenges American model companies.

Strengthens China resolve for domestic reliance

Nvidia’s own Nemotron models help expand AI use on its platforms, giving the company a commercial reason to keep investing in open models. That commercial logic remains in place under the new leadership. Its quarterly filing nevertheless warns that restrictions on supporting Chinese foundation models could harm its business. Political pressure could thus constrain cooperation even where the company still wants it.

For China, a crisis at Nvidia offers little immediate relief from these dependencies. More uncertain access would give developers additional reason to invest in domestic alternatives. Over time, that could help Chinese suppliers gain customers and experience. Building reliable computing clusters and adapting software takes sustained engineering work, however. A faster shift could initially raise costs and slow projects.

The outcome could differ across time horizons. Tighter controls may increase America’s immediate leverage over Chinese AI development. Persistent uncertainty could also encourage Chinese firms to design future systems around alternative platforms, eroding American suppliers’ longer-term influence. The handling of Nvidia’s crisis could affect both processes.

Some direct sales remain possible under controlled conditions. In January, the US Commerce Department introduced case-by-case review for exports of H200 chips and comparable products to China, subject to security and supply requirements. These approvals are conditional.

Tighter control over remote access to restricted technology

Renting computing power from a separate overseas provider raises a different issue. The Remote Access Security Act, passed by the House in January and awaiting Senate action, would expand authority to control remote access to restricted technology.

The distinction matters. One set of controls follows the physical supply of chips and the identity of their recipients. The proposed expansion would restrict access to computing capacity without a chip crossing into China. For developers, the location of a data centre would then offer less assurance about continued access.

There are already signs of concern. It was reported that some Chinese AI companies were seeking to preserve Nvidia-powered capacity overseas because they feared tighter US restrictions. Their behaviour suggests that expectations of policy change can affect investment before new rules take effect.

My larger concern is what this does to the relationship between China and the US. Commercial ties give companies reasons to resist disruption and governments reasons to consider its costs. Their stabilising effect becomes weaker when maintaining contact exposes firms to growing political suspicion.

Damaging feedback could follow. Washington tightens access, Chinese companies reduce dependence, and American firms have less business to defend in China. Each side may then see fewer benefits in preserving the remaining connections. A corporate crisis could add momentum to that process without resolving the underlying security questions.

Commercial ties as a stabilising force

The investigations should establish specific conduct and responsibility. Their findings can inform proportionate responses. Turning disputed allegations into a general judgment about ordinary commercial contact would make it harder to maintain boundaries between enforcing rules and continually expanding them.

Beijing also has an interest in preserving workable routes for compliant trade. Strengthening domestic technology can coexist with continued commercial exchange. If both countries treat every remaining dependency as a vulnerability to eliminate, they will gradually remove the constituencies with a direct stake in avoiding a rupture.

China and the US need arrangements that can withstand this kind of shock. Clear licensing conditions and explanations of policy changes would help companies plan. Government communication should allow specific compliance concerns to be addressed without leaving every legitimate transaction exposed to the next political controversy.

That is one practical requirement for building a constructive relationship of strategic stability between the two countries. Businesses need enough policy continuity to act on high-level commitments to stabilise relations. The handling of Nvidia’s crisis will show whether Washington and Beijing can contain a contested case while preserving predictable space for commercial ties. Their ability to do so will matter long after the immediate shock of Huang’s departure has passed.

Related: Jensen Huang’s affair puts US-China AI chip rivalry on a knife-edge | Jensen Huang: Security risk or witch hunt?