Has the US stoked China-Nvidia feud?

When Washington loosened its restrictions on Nvidia’s exports to China earlier this summer, it looked on the surface like a tentative thaw in a bitter technology standoff. US President Donald Trump authorised the sale of Nvidia’s H20 processor, a chip designed specifically for the Chinese market after earlier export controls had barred the company’s most advanced models.

Yet, instead of welcoming the concession, China seems to have quickly turned against the H20 chips and Nvidia, questioning its safety and urging domestic firms to avoid using it.

_T_he reversal highlighted the volatility of the relationship between the world’s most valuable chipmaker and its second-largest market. It also raised a deeper question: whether the US’s own rhetoric and policy shifts have aggravated, rather than eased, the clash between Nvidia and China.

Washington’s long shadow on Nvidia

The tensions date back to 2022, when the US began imposing restrictions on the export of advanced artificial intelligence chips to China, citing national security concerns. In 2023, the then Biden administration introduced tighter controls, prompting Nvidia to engineer the H20, a “watered-down” model meant to comply with the new thresholds.

Even that chip was caught up in further restrictions as sales were abruptly halted in April 2024 before being re-authorised in July.

Nvidia had lobbied hard for such a reversal. Jensen Huang, the company’s chief executive, personally pressed Trump to loosen restrictions, warning that cutting off China entirely would accelerate the diffusion of rival technologies and weaken US leadership, reports said. The company also argued that its chips posed no national security risks and were designed strictly for commercial applications.

China had likewise called repeatedly for Washington to relax its semiconductor curbs. Beijing regarded the ban as a deliberate attempt to cripple its technological development and had urged more open trade in AI processors. In that sense, the July decision to permit H20 exports looked like a partial success for both Nvidia and Beijing.

Did ‘insult’ change the mood?

But any optimism quickly evaporated after comments from US Commerce Secretary Howard Lutnick. In a television interview, Lutnick remarked that America would never sell China its best chips, not its second-best, “not even our third-best”, a Financial Times report said.

He added that the strategy was to sell just enough to make Chinese developers dependent on the American software ecosystem.

The Financial Times report said several senior Chinese leaders found the comments “insulting”.

Regulators including the Cyberspace Administration of China (CAC), the National Development and Reform Commission (NDRC) and the Ministry of Industry and Information Technology (MIIT) quickly mobilised to discourage domestic firms from purchasing the H20. Informal “window guidance” was issued instructing leading internet firms such as ByteDance and Alibaba to halt orders, while Nvidia executives were summoned over alleged “serious security issues”.

Beijing’s shift showed how rhetoric from Washington can shape China’s regulatory responses. What was supposed to be a US concession became, in Beijing’s eyes, a reminder of America’s intent to maintain dominance.

China’s time to raise security flag

Chinese authorities said they were concerned that Nvidia’s chips could contain risks such as “tracking and positioning” or “remote shutdown” features, allegations the company strongly denied, CNN reported. The CAC even claimed that US experts had raised red flags about potential backdoors, though no public evidence was presented.

By late July, several major Chinese technology companies had either suspended or sharply reduced their H20 orders, reflecting both regulatory pressure and uncertainty over whether future purchases might be sanctioned. Nvidia insisted that the H20 had no military applications and emphasised that both governments recognised it as a commercial product.

Is China looking beyond Nvidia?

Some commentators have said that China’s rejection of the H20 chips forms part of its longer-term strategy of reducing reliance on foreign technology. The South China Morning Post reported that China has mandated its data centres to source more than 50 per cent of their chips from domestic producers by 2025. Officials in Shanghai were among the first to issue such quotas, which have since become national policy.

However, industry experts point out that Chinese chips lag behind Nvidia in certain areas such as memory bandwidth. But China aims to make them increasingly competitive, especially in tasks such as inference, where AI models respond to queries. Huawei, in particular, has made advances in chip development, though analysts said its ecosystem and production capacity remain constrained, the South China Morning Post report said.

Some Chinese policymakers have even pushed for banning foreign chips outright in inference workloads, which represent most AI demand, although such a step appears unlikely in the near term given limited domestic supply, Financial Times said.

This creates a dilemma for Nvidia

For Nvidia, the developments created a dilemma. After receiving US approval, the company had placed large orders for H20 production with TSMC, Samsung and other suppliers. But with China now holding back, Nvidia has reportedly asked some partners to suspend production and has shifted to managing its existing inventory.

Huang publicly maintained that shipping H20 chips to China posed no national security risks and expressed gratitude for Washington’s approval. But he admitted that Nvidia’s fate in the Chinese market ultimately depended not on demand but on political decisions in both capitals.

Meanwhile, reports suggest Nvidia is already working on a new China-specific chip, tentatively called the B30A, based on its latest Blackwell architecture. That model would be more powerful than the H20 yet still fall below the thresholds of US export restrictions. Whether Washington would permit such sales remains unclear, and Chinese regulators’ scepticism about US chips may persist regardless, Reuters said.

On the other hand, a Bloomberg report said that Nvidia has instructed parts suppliers including Samsung Electronics and Amkor Technology to stop production related to H20 AI chips.

Is China sending dual message?

The contrasting signals from Beijing’s regulators also revealed a deeper ambivalence. The Financial Times reported that ministries in charge of trade and foreign affairs welcomed Huang’s visit to China in July, seeing it as a chance to showcase goodwill and keep open channels of cooperation. But agencies like the CAC and NDRC pressed for reduced reliance on Nvidia and more aggressive adoption of Chinese chips.

This dual message reflected the tension between short-term needs and long-term goals. China’s AI developers still rely heavily on Nvidia’s CUDA software ecosystem, which makes switching costly and technically complex. Yet policymakers see domestic substitution as essential to strategic independence.

What is the US’s role in the rift?

Taken together, the episode illustrates how the tussle between Nvidia and China has been shaped as much by the US policy as by market forces. Export controls created the H20 in the first place, while Washington’s rhetoric and enforcement strategies helped push Beijing to reject it. What began as a commercial negotiation over chip specifications has escalated into a symbol of national pride and strategic autonomy.

For Nvidia, the result has been a collapse of demand for its China-specific chip and uncertainty about future approvals. China, on the other hand, seems to be using this as an opportunity to boost domestic chip industry. The US, under President Donald Trump, has been vocal about domestic bases for American companies and signalling that the US-based companies should align with the administration’s broad policies. The result is a tug of war, possibly at the expense of a tech giant’s business prospects.