Nvidia, directed by its founder and CEO Jensen Huang, has become a key piece of the geopolitical board between the United States and China. Decisions about its chips – particularly the H20 model designed for the Chinese market – are loaded with political, economic and strategic consequences.
First, there is the CUDA standard. Nvidia developed this programming ecosystem that, although invisible to the general public, is essential for any company to work. Accepting that standard means depending on American technology. Washington understood that if China uses any CUDA compatible chip, it is tied to that system. That motivated Donald Trump’s government to allow the sale of H20 for six months: a “limited” chip that meets restrictions, but forces China to use CUDA and follow the technical ecosystem imposed by the US.
Second, rare earths. China controls the global production of these critical minerals for the manufacture of magnets, electric cars, electronics and defense. During the technological conflict, Beijing suspended exports in response to US restrictions on Nvidia chips. It was a strategic play: without access to H20, there would be no rare earth supply. Finally, to avoid a global economic chaos, the US raised part of the veto and resumed the sales of H20.
Third, the Chinese accusation. After authorizing the sale of the H20, the Chinese cybersecurity authority of China demanded public explanations to Nvidia. They affirmed that the H20 could trace location and deactivate remotely – although they did not present technical audits to prove it. It was a clear complaint: a message and a political threat, not a technical debate. In this context, China seeks to project power, show that it does not yield to the United States and warn countries that aspire to develop sovereign artificial intelligence (AI): “Do not accept chips that can be monitored.”
Fourth, the problem of espionage and sovereignty. China states that a chip with tracking functions would be unacceptable for countries that wish to have an independent technical infrastructure. In addition, without Nvidia chips – and without CUDA – there is no technological basis to train or execute advanced AI systems. Although it is avoided by talking directly to artificial intelligence, the background is political: no country can build a competitive technological infrastructure if it depends on standards imposed by others.
Fifth, in Chinese media resurfaced grandiloquent statements according to which Huawei chips exceed those of Nvidia. However, there is no concrete technical tests or comparisons. It is part of a nationalist narrative: “China lies, but its statements are strong; few real events.” This strategy seeks to erode international confidence in NVIDIA and reinforce the local industry, although it still does not have the real capacity to compete, both in performance and in technical ecosystems.
Sixth, all this happens just before an important meeting of the Communist Party, where Xi Jinping needs to show firmness. The economy grows less, there is youth unemployment and internal political pressure. Publicly accusing an American technological superpower serves to save the “face”: transmit dignity, sovereignty and control. Being perceived as weak against Washington would be dangerous. That is why China acts where it can: project power, even without technical evidence.
On the American side, the argument was clear: it was necessary to ensure that China consumes more American ecosystem technology. Allowing H20 was a controlled play: they maintain technical influence without delivering their most powerful chips. Pressures on Nvidia also came from congressmen and security experts. Some even promote laws that will demand that chips include physical trackers to avoid contrabing sanctioned countries. That is, USA also evaluates control mechanisms similar to those now denounced in China.
Meanwhile, Jensen Huang traveled to Beijing after the lifting of restrictions. His mission was diplomatic. He sought to assure Chinese clients and authorities that the company is reliable, without abandoning its base in Washington. Nvidia is trapped in a dilemma: if it demonstrates too much loyalty to China, it loses support in the US; If you move too much, you lose your access to the Chinese market.
Finally, on this board there are no clear winners. China needs chips and standard; USA needs to maintain Chinese dependence. Beijing tries to replace imported technology with Huawei, Biren or Cambricon, but there is still nothing equivalent to CUDA or an ecosystem that replaces it. At the same time, the strategy of accusing without technical evidences continues: strong words, weak facts.
This conflict is not a chip or technology in abstract. It is a battle for sovereignty: who defines the standards, who controls raw materials and who imposes the rules of global technical power. Nvidia is in the center, and each move – the dispatch of chips, a public threat, an executive visit – is part of a game where the hardware capabilities not only matter, but which country marks the architecture of the future.
Things as they are.
Image gallery
