US AI exports to Asia face cheaper Chinese model push
Washington is promoting American AI tools across Asia as Chinese firms offer lower-cost models and Beijing courts regional governments.
By Jordan Bell · Startups & Deals Reporter
· 4 min read
US AI exports to Asia are becoming a sharper test of Washington’s tech strategy as Chinese companies sell lower-cost alternatives and Beijing promotes its own systems across the region. For retail investors, the fight matters because it touches the same companies driving AI spending: chipmakers, cloud providers and model developers.
The U.S. has introduced an American AI Exports Program and other efforts meant to encourage foreign buyers to use U.S. technology, according to CNBC. Gary Dvorchak, managing director at The Blueshirt Group, told CNBC that Washington’s goal is to prevent China from becoming the main AI supplier for Asia and other global markets, while noting that U.S. companies still offer a broader package that can include chips, infrastructure and models.
At the Asia-Pacific Economic Cooperation Digital Weeks event in Chengdu, China, that sales effort appeared quieter than it was at last year’s first APEC AI meeting in South Korea, CNBC reported. In 2025, Michael Kratsios, President Donald Trump’s chief science and technology policy adviser, promoted the U.S. AI Action Plan and the American AI Exports Program, according to a White House transcript.
Why is China gaining ground in AI across Asia?
Chinese AI firms have been releasing models with capabilities similar to U.S. systems at much lower prices, according to CNBC. Many Chinese models are also open-source or open-weight, meaning developers can inspect, adapt or run parts of the model more freely than they can with many closed U.S. systems.
That cost difference matters in emerging Asian economies, where governments and businesses may want local control over technology and lower deployment costs. Wei Sun, principal analyst for artificial intelligence at Counterpoint Research, told CNBC that APEC’s support for open-source AI with strong security gives China’s strategy more regional legitimacy.
China used its role as APEC host to highlight that approach. Chinese President Xi Jinping said at the World AI Conference in Shanghai on July 17 that China would provide developing countries with 5,000 AI training and seminar opportunities and work on AI application cooperation centers with Southeast Asia and other regions, according to Chinese state information cited by CNBC.
China also sent Vice Premier Zhang Guoqing to the APEC Digital Weeks ministerial event on July 23 to call for technology standards developed with Asia-Pacific countries, CNBC reported. Later that day, APEC’s 21 member economies, including the U.S., agreed to support open-source AI with strong security.
What is the U.S. offering?
Bill Guidera, deputy under secretary for innovation and engagement at the U.S. Department of Commerce, spoke at a July 24 APEC forum organized by China’s cybersecurity regulator, according to materials reviewed by CNBC. He promoted the exports program and said buyers could purchase either a complete U.S. technology stack or selected pieces of it.
The Commerce Department’s International Trade Administration confirmed Guidera’s Chengdu appearance in a July 29 social media post. Asked about a Politico report that the program had received 78 applications, fewer than expected according to three former officials cited by Politico, an ITA spokesperson told CNBC that application volume exceeded expectations. The White House did not respond to CNBC’s request for comment.
U.S. corporate visibility in Chengdu was limited, CNBC reported. Google and Meta were the only U.S. companies CNBC observed with booths as of July 23. Google focused on AlphaFold, its molecular AI system, and Meta highlighted AI tools for small businesses. Google government affairs vice president Wilson L. White made only brief references to Gemini in a July 24 speech, while Tencent Vice President Cai Guangzhong spoke afterward about adoption of Tencent’s Hunyuan large language model and a cloud project in Thailand.
Analysts do not see Asia as a clean split between U.S. and Chinese systems. Sun expects more mixing of technologies, and Votee AI CEO Pak-Sun Ting told CNBC that Southeast Asian users often rely on Nvidia chips for AI training while using other chips for running models. He said Votee’s Cantonese open-source model was developed partly with Alibaba’s Qwen model.
Yue Su, principal economist at the Economist Intelligence Unit, told CNBC that the U.S. and China are competing through different AI strategies but cannot fully separate from each other. She also said the smaller U.S. presence at APEC was not surprising given other events, including a July 24 AI summit in San Francisco attended by South Korea’s president and U.S. AI leaders Sam Altman, Dario Amodei and Jensen Huang.
This story draws on original reporting from CNBC.
