US Seeks Asian AI Adoption Amid China’s Dominance in Cheaper Models

BEIJING – The intensifying artificial intelligence rivalry between the United States and China is playing out across the vast continent of Asia, a critical theater for global technological dominance.

“The United States’ strategic objective is clear: to prevent China from establishing itself as the leading AI supplier for the rest of Asia, and by extension, the entire world,” observed Gary Dvorchak, managing director at The Blueshirt Group. He further elaborated that while China often presents more cost-effective AI alternatives, the U.S. currently offers a more vertically integrated solution, encompassing everything from cutting-edge semiconductors to sophisticated AI models.

This U.S. sales challenge, however, was subtly underscored during the recent “Digital Weeks” event hosted by the Asia-Pacific Economic Cooperation (APEC) in the southwestern Chinese city of Chengdu. The American presence was notably understated, with few prominent public displays of engagement, despite earlier efforts by a U.S. official and a U.S. business representative to APEC to highlight the event’s focus on AI.

This subdued participation by the U.S. comes amidst a period of significant flux. Notably, Anthropic experienced a highly publicized pivot regarding the release of its Fable AI model, reportedly due to abrupt shifts in U.S. policy. Concurrently, China has been rapidly deploying new AI models, such as those from Kimi, that offer comparable capabilities to leading Western counterparts at a fraction of the cost, intensifying competition and raising new strategic considerations for the U.S.

In stark contrast, at the inaugural APEC AI meeting held in South Korea last summer, Michael Kratsios, then-chief science and technology policy advisor to President Donald Trump, robustly championed the U.S. AI Action Plan and the establishment of the American AI Exports Program. This initiative was designed to foster international partnerships for U.S. AI technologies.

However, recent reports suggest a less-than-robust uptake for the American AI Exports Program. A Politico report, citing three former officials, indicated that the Commerce Department had received a considerably lower number of applications than anticipated. This follows earlier remarks from a U.S. Department of Commerce official, Bill Guidera, at an APEC High-level Forum on AI organized by China’s cybersecurity regulator. While Guidera reiterated the program’s goals and highlighted the “brilliant design” of U.S. AI, emphasizing its strength and security, the program’s overall reception appears to be a work in progress.

In response to questions about the application numbers, an International Trade Administration spokesperson stated that the volume of applications “exceeded our expectations,” a sentiment not directly corroborated by independent reporting. The White House did not provide a comment on the matter.

Commercial exhibitions at the event further highlighted the U.S. presence’s limited scope. Google and Meta were among the few American companies observed with booths, focusing on specialized AI applications like Google’s AlphaFold for molecular AI systems and Meta’s AI tools for small businesses, rather than showcasing their most advanced large language models. Google’s government affairs vice president offered only brief mentions of Gemini during his remarks. In contrast, Tencent’s Vice President took the stage immediately afterward to prominently detail the expanding adoption of its Hunyuan LLM and its cloud initiatives in Thailand, underscoring China’s aggressive push into regional markets.

As the host of APEC this year, China is leveraging the platform to amplify its AI capabilities during a period of heightened U.S.-China geopolitical and technological tensions. Beijing’s strategy emphasizes its predominantly open-source AI models, contrasting with the largely proprietary nature of many U.S. offerings.

Chinese President Xi Jinping announced at the World AI Conference in Shanghai that China would offer 5,000 AI training and seminar opportunities to developing nations. Furthermore, plans are underway to establish AI application cooperation centers with Southeast Asia and other regions, signaling a clear intent to build an extensive AI ecosystem under Beijing’s influence.

The engagement continued with Vice Premier Zhang Guoqing advocating for the development of robust tech standards with other Asia Pacific nations at the ministerial-level APEC Digital Weeks. This diplomatic push culminated in an agreement from the 21 member economies, including the U.S., to endorse open-source AI with “strong security.”

“The endorsement of open-source models with strong security assurance lends greater regional legitimacy to China’s open-weight strategy, particularly across emerging Asian economies where deployment cost and technological sovereignty are crucial decision-making factors,” explained Wei Sun, principal analyst for artificial intelligence at Counterpoint Research. This strategic move positions China favorably in markets prioritizing accessibility and local control over proprietary solutions.

Rather than envisioning a complete bifurcation of the AI landscape into distinct U.S. and Chinese spheres, analysts anticipate a more nuanced reality characterized by technological convergence, especially within Asia.

The sheer linguistic diversity across Asia, with over 1,300 living languages in Southeast Asia alone, presents a significant challenge for a one-size-fits-all AI model approach. This complexity necessitates the development of AI systems specifically tailored to local languages and cultural contexts.

Governments across Asia and beyond are reportedly investing “billions” in bespoke AI solutions designed to address these linguistic nuances. According to privately funded startup Votee AI, its CEO, Pak-Sun Ting, is collaborating with at least five governments, including two in Southeast Asia, and the company is already generating substantial revenue exceeding $10 million annually.

Ting noted a trend in Southeast Asia: while Nvidia chips are frequently favored for AI training, other chip architectures may be employed for model inference. He highlighted Votee’s open-source model for Cantonese speakers, which was partially developed using Alibaba’s open-source Qwen model, illustrating the collaborative and hybrid approach emerging in the region.

Fundamentally, the ability of AI to generate tangible economic returns remains the paramount consideration, irrespective of its origin. “While the U.S. and China are engaged in an intense competition in AI technology and diplomacy through their distinct approaches, complete decoupling between them is ultimately unfeasible,” stated Yue Su, principal economist at the Economist Intelligence Unit.

Su views the limited U.S. presence at APEC as unsurprising, given other concurrent high-profile events such as a significant AI Summit in San Francisco. This summit, organized by South Korea’s tech ministry, provided an opportunity for President Lee Jae-myung to engage with leaders of frontier AI models, including Sam Altman of OpenAI, Dario Amodei of Anthropic, and Jensen Huang of Nvidia. This engagement aimed to build upon South Korea’s substantial investments in chip and AI megaprojects.