Tweet

The indecision surrounding US export controls on AI chips to China highlights the Trump administration’s erratic and inconsistent tech policy, casting the American-led strategy of decoupling into doubt, even among its own allies.

Decoupling from China has served as one of the fundamental tenets of United States (US) tech and export policy for almost a decade, particularly in light of the growing strategic competition between the two nations. Given its increasing importance and integration across virtually all existing technologies, Artificial Intelligence (AI) has been a primary instigator in this regard. This has rendered AI export policy particularly significant. However, the Trump administration’s decision to potentially revert export bans on the sale of AI chips to China cast America’s existing stance on the issue into serious doubt.

In light of China’s retort and its recent decision to rely on domestically manufactured chips for its data centres, the Trump administration eventually decided to ban the sale of Nvidia’s newly developed B30A chips to China. The whole episode highlights Trump’s confused and meandering tech policy. In addition to raising domestic opposition, it has become a cause of concern for US allies. The grand strategy of decoupling technology and critical mineral supply chains from China, initiated by the US, has had a sizeable — and, in many cases, detrimental — impact on multiple economies around the world. This has prompted many to question the efficacy of the strategy and the manner of its execution in the future.

US AI Policy and Strategic Initiatives

The Trump administration has announced a slew of AI and tech policy measures in recent months, including the Stargate Project, America’s AI Action Plan, as well as the extension of tax credit under the CHIPS Act. All these initiatives share the common goal of boosting domestic manufacturing, reducing dependence on Chinese manufacturing, and inhibiting China’s progress in critical technologies through export restrictions.

As a case in point, Nvidia’s H20 chips, a downgraded version of the company’s most advanced chips (such as the Blackwell) and specifically tailored to cater to the Chinese market, were developed to comply with export controls put in place during the Biden administration. In April 2025, Trump placed a ban on the sale of H20 chips to China, which ended up causing a charge of about US$4.5 billion for Nvidia in the first quarter of the year. According to CEO Jensen Huang, the ban is further expected to cost the company about US$8 billion in the July quarter.

The US-China Tug of War on AI Chips

In August 2025, Trump relayed his intention to reverse the previous ban on Nvidia and AMD chips in return for a 15 percent sales revenue cut to the US government. In addition to being potentially construed as unconstitutional export taxes, the move also raised questions over the American policy of decoupling from China. While there are multiple opinions on the motive behind the decision, a few stand out in particular.

The Trump administration’s stance on AI policy vis-à-vis China is hypocritical and conflicting. Countries such as Taiwan, Japan, and South Korea are being forced to relocate chip manufacturing facilities outside China due to US export controls, incurring significant financial losses in the process. Meanwhile, America’s own wavering stance on AI chip exports to China does little to inspire confidence in the overall strategy.

The announcement followed months of lobbying by Nvidia, which subsequently led to a bilateral meeting between Trump and Huang in July. Despite being one of the primary instigators of decoupling from China, Trump’s apparent decision to succumb to Nvidia’s demands seemed to indicate a conflation of strategic and corporate interests. The justification he offered was that H20 chips were a downgraded and inferior version of the cutting-edge Blackwell chips. However, this did not justify the sale of a strategically important technology, particularly given China’s propensity for reverse-engineering Western tech. Moreover, the option of providing remote access through cloud services was a viable alternative to selling physical hardware to the Chinese.

Another possible motive may have been the pursuit of a US-China trade deal, with the move intended to lull the latter into a false sense of security through the sale of outdated chips. However, the strategy was quickly thwarted by China, which banned its companies from purchasing H20 chips in September 2025. This decision likely stemmed from the announcement of potential US Congress legislation mandating tracking and kill-switch capabilities in exported chips, alongside comments by the US Secretary of Commerce, Howard Lutnick, on a “strategy of addicting the Chinese” to obsolete technology, among other issues. China doubled down on its decision by further ordering its data centres that are less than 30 percent complete to remove all installed foreign chips or cancel plans to purchase them. In response, the White House announced its decision to ban the sale of Nvidia’s scaled-down B30A chips to China on November 6, 2025.

The primary line of reasoning being used to justify Washington’s recent decision is opposition from China hawks such as Secretary of State Marco Rubio, who cited national security concerns related to supplying AI chips to China. These concerns stem from the potential military implications of the technology’s dual-use nature. However, as noted earlier, Trump was open to the idea until only a few months ago. Based on the timing of the decision, the more likely reason appears to be the signing of a US-China trade deal just a few days earlier on the sidelines of the APEC Summit in South Korea. This was compounded by China’s continued commitment to reducing its reliance on foreign chips.

Trump’s AI and Tech Policy: Geopolitical Fallout

While the transactional nature of Trump’s policies is well known, it is not always a viable strategy when it comes to long-term foreign policy, which is dictated by the needs of partners and allies in addition to serving national interests. In this context, the Trump administration’s stance on AI policy vis-à-vis China is hypocritical and conflicting. Countries such as Taiwan, Japan, and South Korea are being forced to relocate chip manufacturing facilities outside China due to US export controls, incurring significant financial losses in the process. Meanwhile, America’s own wavering stance on AI chip exports to China does little to inspire confidence in the overall strategy.

The lack of a consistent US tech policy is creating uncertainty and confusion around the Western-led policy of decoupling. It is also adding credence to the idea of a fledgling American hegemony, prompting US allies to question their allegiances.

If Trump continues with unilateral decision-making, it could lead to further resentment against the US, even among its closest allies. This may prompt them to forge new alliances in order to gradually reduce their dependence on American technology supply chains. An example of this already seems to be materialising in growing India-Russia-China engagement following the 25^th Shanghai Cooperation Organisation (SCO) Summit at Tianjin. The situation is being further exacerbated by India’s insistence on maintaining a cordial relationship with Russia despite multiple threats issued by the Trump administration.

The lack of a consistent US tech policy is creating uncertainty and confusion around the Western-led policy of decoupling. It is also adding credence to the idea of a fledgling American hegemony, prompting US allies to question their allegiances. The absence of firm and predictable leadership in Washington, coupled with growing uncertainty around Trump’s policy on critical issues like tech and AI, is pushing other nations towards pursuing new strategies and alliances independent of the US. This may spell disaster for the country in the years to come.

Prateek Tripathi (Centre for Security, Strategy and Technology)
12 December 2025

 

Tweet