The Jensen Huang scandal could have an unexpected consequence for the technology war between Washington and Beijing. Fearing tougher US restrictions, several Chinese companies are urgently seeking to secure access to Nvidia processors hosted abroad. A new escalation that could paradoxically accelerate China's technological independence.
"Watch the fire burn from the other shore." This ancient precept, drawn from the 36 Chinese stratagems, could well sum up Beijing's position in the face of Jensen Huang's troubles. As Nvidia faces an unprecedented governance crisis, China could take advantage of the difficulties confronting its main supplier of artificial intelligence chips to accelerate its technological independence.
According to Reuters, several Chinese AI companies are nevertheless attempting to urgently reserve computing power at foreign data centers. They fear that revelations surrounding Nvidia's CEO could prompt Washington to close one of the last remaining avenues for accessing the most advanced US processors.
For several years, the United States has sought to slow China's progress in AI by limiting semiconductor exports. But each new restriction also strengthens Beijing's arguments in favor of a domestic industry capable of doing without US technologies.
Washington wants to close the last avenue for accessing Nvidia chips
If Washington bans the export of certain Nvidia chips to China, Chinese companies still have a way to access them. Under certain conditions, they simply need to rent computing power from data centers located in Singapore, Malaysia or elsewhere around the world. The processors remain physically abroad, but Chinese companies can use them remotely to train their artificial intelligence models.
This is precisely the back door that Washington is considering closing. The Remote Access Security Act, passed by the House of Representatives in January but still awaiting consideration by the Senate, would give US authorities more tools to regulate these services. In other words, the United States could restrict not only chip exports, but also their use by Chinese companies, even when the chips are installed in foreign data centers.
And the Jensen Huang affair could strengthen the arguments of its supporters.
The executive acknowledged an affair with Alice Huang, a businesswoman linked to Megaspeed, a Singapore-based company specializing in computing infrastructure. According to The New York Times, US intelligence agencies suspect the latter of having ties to Chinese intelligence, allegations disputed by Beijing and not publicly established.
Megaspeed is also facing questions over Chinese companies' access to Nvidia processors subject to US restrictions. Two US senators have specifically cited more than $4.8bn in advanced chips purchased by the company since June 2024.
For Washington, the stakes therefore go beyond the physical destination of the semiconductors. The issue now is controlling who is actually using their computing power.
Huawei, the long-term winner?
In the short term, tighter restrictions would penalize Chinese developers, who remain heavily dependent on the most advanced US technologies.
But they could also accelerate a transition that is already underway. According to IDC, cited by Reuters, Nvidia's share of China's data center AI accelerator market fell to 34% in the first half of 2026, with shipments down 44% year over year.
Huawei appears to be the main beneficiary of this reshuffling. The group is developing its Ascend processors, while Cambricon, MetaX and other domestic players are seeking to gain ground.
These companies still face significant industrial limitations, particularly in manufacturing advanced chips and securing supplies of HBM memory. But US restrictions are gradually giving them a captive market, in which Chinese customers are being encouraged to adapt their software and infrastructure to domestic technologies.
This is the paradox at the heart of the US strategy. By seeking to prevent China from accessing the best processors, Washington could accelerate the emergence of a competing ecosystem that is less dependent on Nvidia.
Nvidia risks losing more than sales
For Nvidia, the immediate commercial impact of another tightening of restrictions remains difficult to quantify. The group does not disclose revenue indirectly tied to Chinese companies using its processors in foreign data centers.
The long-term risk is more significant. As Chinese developers migrate to domestic architectures, they could gradually move away from CUDA, the software environment that represents one of Nvidia's main competitive advantages.
The development is all the more ironic because Jensen Huang has long argued that allowing Chinese companies to use US technologies also helps preserve America's technological influence.
China cannot yet do without the most advanced US chips. But by repeatedly closing off access to them, Washington could accelerate the construction of an AI market that tomorrow will no longer need Nvidia.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.