The United States and China are planning to hold talks on artificial intelligence (AI) in September, Reuters reported on Tuesday (the 21st), citing five sources who requested anonymity. This is part of discussions between the two superpowers on how to regulate the risks posed by increasingly powerful, competitive cutting-edge technology models.

The AI talks, a key outcome of the Trump-Xi summit last May, are likely to take place before President Xi Jinping's visit to the U.S. on September 24, although a specific date has not yet been set, according to the sources.

According to four of the sources, Treasury Secretary Scott Bessent is expected to lead the talks for the U.S. side. Three of them stated that other participants, the agenda, and the venue for the U.S.-China talks are still under discussion, and that the preparations are in the early stages.

Media reports indicate that the current U.S.-China relationship in the AI industry has moved away from the past dynamic, where the U.S. led and China followed. Now, both countries are engaged in an all-out battle, with distinct strategic orientations as they divide the global AI market.

Recently, Chinese startups such as Moonshot AI with its "Kimi K3" and DeepSeek have released AI models capable of performing complex tasks at costs significantly lower than those of major U.S. tech models. Reports have also emerged that China has begun to outpace the U.S. in quantitative research achievements, such as the number of published papers and patent applications.

However, while the U.S. and China may appear similar in their quest for AI supremacy, they are difficult to compare given their differing objectives.

The U.S. is focusing on the development of Artificial General Intelligence (AGI) or super-intelligence models that surpass human cognitive abilities, led by private tech giants like OpenAI, Google, and Anthropic. This structure leverages vast capital and energy resources to drive innovation in foundational technologies.

Conversely, amidst U.S. semiconductor regulations, China is realistically going "all-in" on profitability and practicality rather than AGI. It is evaluated to be focusing on the rapid integration of AI infrastructure into traditional industrial settings—such as manufacturing plants, smart cities, ports, and finance—as well as expanding into physical AI (robotics, etc.).

Furthermore, the U.S. government views AI technology as a core pillar of national security and is mobilizing all available means to block China's growth.

As a result, the dominant analysis is that U.S.-China AI relations have entered a stage of "decoupling," completely severing mutual technological dependence. It is argued that they are locked in an uncompromising, all-out war: the U.S. to protect its lead in foundational technology and infrastructure, and China to secure independent cost-efficiency and a vast market for industrial application.

The problem is that South Korea’s AI and tech industries are trapped in a massive "sandwich" crisis. Domestic tech firms like Naver, Kakao, and KT warn that if they become dependent on the AI ecosystems of the U.S. (OpenAI, Google) or China, they risk losing national data sovereignty and cultural identity.

Chinese firms, led by DeepSeek, offer high-performance AI APIs at one-hundredth the cost of U.S. alternatives, posing a major threat to price competitiveness. Simultaneously, as the U.S. drastically tightens export controls on advanced AI semiconductors (such as HBM) to China, companies like Samsung Electronics and SK Hynix face pressure to abandon the massive Chinese market.

As South Korean semiconductor firms must integrate into the next-generation AI memory supply chain prioritized by the U.S., there are growing calls to find a "third alternative." This would involve targeting countries in the Middle East (Saudi Arabia, UAE), Southeast Asia, and Europe—which find it burdensome to rely solely on either U.S. or Chinese AI technology—and making the "construction of customized AI infrastructure" a new engine for exports.

U.S. NNP = Lead Reporter Hong Sung-gu / Special Syndicate for this journal, NNP info@newsandpost.com