Conceptual diagram of AI Photo: VCG

The US is reportedly pressuring dozens of countries to take sides in the global AI competition with China, according to a leaked US State Department draft letter seen by Reuters. The coercive alignment requirement, Chinese observers said, marks a typical zero-sum geopolitical maneuver that may split global AI industrial and supply chains and undermine multilateral technological collaboration. As AI development calls for multipolarity, the US, by forcing countries to take sides, is embarking on a futile geopolitical gamble that is doomed to fail.

Unlike the US' unilateral approach, China has launched a series of global AI cooperation initiatives centered on openness, practical cooperation and shared development, an approach that could help ensure AI advances benefit a broader range of developers, industries and consumers worldwide, industry analysts said.

Coercive division

The US is preparing to ‌tell dozens of countries they must pick sides in the artificial intelligence race with China, claiming that they will be excluded from a US-led AI coalition if they also sign up for China's AI framework, according to a US official and an internal draft reviewed by Reuters.

The draft letter, prepared by the State Department, is addressed to the 35 signatories of a US "AI Opportunity Statement" signed in June, which includes members of the non-binding Pax Silica framework and other countries that have expressed a desire to align cooperation on AI with Washington, Reuters reported.

"To be part of everything is to be part of nothing. Signature of the Pax Silica Declaration is not merely a membership subscription, but a commitment," the letter says, urging countries to "choose deliberately" on AI.

"It cannot be held alongside membership in duplicative initiatives whose expectations conflict with our own," the letter said, without specifically mentioning China.

Reuters said it could not determine when the US intends to send the letter or whether it might be amended before sending. The draft was undated.

Launched by the US last year, Pax Silica focuses on consolidating Western dominance over AI model supply chains, semiconductor production and critical mineral resources. Nearly two dozen countries have joined the mechanism, including Japan, South Korea and Australia, according to Reuters.

US officials drafted the letter to make clear that "you can't have it both ways," the US official told Reuters, speaking on condition of anonymity given ongoing internal discussions on the issue.

Washington's move to pressure some countries into taking sides in the global AI race lays bare its deep-seated anxiety over China-US tech competition, Li Yong, an executive council member of the China Society for WTO Studies, told the Global Times on Sunday.

As China's open-source AI models achieve rapid breakthroughs and advance AI as a global public good, Washington is attempting to politicize technological cooperation and weaponize supply-chain rules to suppress China's AI development, while forcing countries into binary choices that run against the common interests of most economies, Li said.

The US wants to cut China off from critical resources and hardware supplies through alignment demands. "Essentially, Washington wants a divided AI world by drawing a digital curtain to split the global market in two, and gain a competitive edge in AI via geopolitical containment," Li said. "This is nothing but a political calculation. It cannot reverse the irreversible trend of advancement for artificial intelligence as a cutting-edge technology."

The US' unilateral moves have put some of its major companies in a bind. The Trump administration does not want Apple to buy memory chips from China, Commerce Secretary Howard Lutnick said in an interview, the Wall Street Journal reported. 

According to the report, insatiable demand for computer memory among companies building AI data centers has led to a shortage of the critical component and skyrocketing prices for what is available. To help alleviate the supply crunch, Apple is reportedly looking to Chinese manufacturers.

Apple's chief operating officer, Sabih Khan, declined to confirm that Apple was testing Chinese memory chips in an interview Thursday. But he said that given the magnitude of the supply shortage, "we have to look at all options," the Wall Street Journal reported.

Apple has been testing memory chips made by Chinese companies CXMT and Yangtze Memory Technologies, a prelude to potentially including them in Apple devices sold in China, the Wall Street Journal reported, citing people familiar with its efforts.

No country wants a semiconductor ecosystem dictated entirely by the US, in which all nations have to accept Washington's high‑handed rules. Instead, AI development demands a multipolar world where countries are granted room for independent choices and growth. This preference for multipolarity means Washington's take‑it‑or‑leave‑it demand is doomed to be a futile geopolitical gamble, Li said.

Experts noted that the US push to build a supply-chain alliance excluding China runs counter to the interests of most countries, and that the prospect of a bipolar AI world divided into two rival blocs would similarly harm the broader international community.

"The Washington-driven initiatives may create two separate sets of AI standards, application ecosystems and geopolitical camps. This directly contradicts the fundamental objective of making AI deliver benefits for all humanity," Li said. "Such coercive tactics merely serve Washington's unilateral interests, stifle global AI advancement and shift the resulting costs onto the entire world."

The US' unilateral approach carries risks and would be difficult to implement, while significantly increasing supply chain costs, Hu Qimu, a deputy secretary-general of the Digital-Real Economies Integration Forum 50, told the Global Times on Sunday.

"The Chinese and US economies are inherently complementary. The US is investing heavily in algorithm development, but these technologies ultimately need real economy and industrial applications to achieve commercialization," Hu said, noting that the AI industry will ultimately need real economy applications to monetize its technologies, indicating that China's vast manufacturing sector and diverse industrial application scenarios remain highly important to the development and commercialization of AI technologies.

Inclusive path

In contrast to the US' unilateral approach, China has rolled out a series of global AI cooperation initiatives aimed at promoting openness, practical cooperation and shared development.

Twenty-nine countries signed an agreement in Shanghai in July to establish the World Artificial Intelligence Cooperation Organization (WAICO). It aims to promote international cooperation and global governance on AI, ensuring that AI is beneficial, safe and fair, thereby promoting its healthy and orderly development to benefit all humanity.

In the next five years, China will provide developing countries with 5,000 opportunities in AI training and seminar programs. China will develop international AI application cooperation centers with the Association of Southeast Asian Nations, the League of Arab States, the African Union, the Community of Latin American and Caribbean States, the Shanghai Cooperation Organization, and BRICS. China will enable 30 countries to use the AI-powered meteorological warning system, or MAZU, to safeguard homes around the world.

"It shows that China stands ready to work with all countries to help less‑advanced AI regions catch up. We will share our extensive practical experience in AI applications via open‑source models, so AI technologies can benefit more nations," Li said. "In the long run, this path will win global recognition, as it prioritizes the well‑being of all humanity, instead of merely pursuing victory in AI competition for a single country."

Chinese AI models are gaining strong traction among users on OpenRouter, an AI model aggregation platform, with models developed by Chinese companies taking four of the top five spots on the platform's LLM leaderboard, according to the latest weekly ranking seen by the Global Times on Sunday. DeepSeek V4 Flash 0731 ranked first with 11.1 trillion tokens, followed by Tencent's Hy3 with 10.4 trillion tokens. OpenAI's GPT-5.6 Luna ranked third, while DeepSeek V4 Flash 0423 and Z.ai's GLM 5.2 ranked fourth and fifth.

The strong performance of Chinese models reflects, in part, their open model strategy, which differs from the approach of many major US AI companies that mainly provide access to their leading models through closed platforms and application programming interfaces, Ma Jihua, a veteran industry analyst, told the Global Times, suggesting that openness and cooperation, rather than exclusion and confrontation, are key to ensuring that AI advances benefit a broader range of developers, industries and consumers worldwide.