The US-China summit held in Beijing in May 2026 appeared calm on the surface. President Donald Trump and Chinese President Xi Jinping projected an image focused on management rather than confrontation. Immediately after the meeting, both nations issued a generic statement expressing their commitment to continue dialogue, which provided some relief to the markets.
However, the essence of this summit was far from a mere diplomatic event. It marked the official beginning of a new era of hegemonic competition over artificial intelligence (AI), semiconductors, energy, manufacturing, supply chains, data, and platforms. Just as the United States and the Soviet Union faced off with nuclear weapons during the Cold War, the 21st century has seen the US and China begin to clash over AI and semiconductors.
One of the most symbolic moments of the summit was when President Trump unexpectedly invited Jensen Huang, CEO of Nvidia, to board Air Force One on his way to China. This was not just a trivial incident; it symbolized that the US prioritized AI and semiconductors in this meeting. While prominent figures from the US tech industry, such as Elon Musk and Tim Cook, were already part of the economic delegation, the urgent inclusion of Nvidia's CEO highlighted that AI semiconductors have become a national strategy.
In reality, the core of the AI hegemonic competition hinges on who can secure more AI chips, build larger AI data centers (AIDCs), and provide more extensive data training and inference services. AI is no longer just a software industry; it has evolved into a massive industrial revolution platform encompassing power, semiconductors, cooling systems, networks, cloud computing, and manufacturing automation.
During the summit, President Xi remarked that “the world has reached a new crossroads,” referencing the so-called 'Thucydides Trap.' This was not merely a historical reference but a strategic warning to the US. China suggested that if the US continues to exert military, economic, and technological pressure, both nations could ultimately be drawn into a vortex of significant conflict.
At the heart of this is what China refers to as the '3T': Taiwan, Trade, and Technology.
First, regarding Taiwan, China reaffirmed its stance of not backing down. The message was clear: if the US deepens its involvement in Taiwan or attempts military intervention, it could serve as a direct trigger for US-China conflict.
Second is trade. Since the first Trump administration, the US has pressured China through tariffs and sanctions. However, China is no longer just the 'world's factory.' In sectors such as electric vehicles, batteries, solar energy, drones, telecommunications equipment, and certain AI fields, it has begun to establish world-class competitiveness. The more the US tries to push China out of the market, the more American companies risk losing access to the Chinese market.
Third is technology, which is the core of this summit. The US has attempted to slow China's AI development through export restrictions on semiconductor equipment and AI chips. However, China is pursuing self-sufficiency much faster than expected.
A prime example is Huawei. Despite strong sanctions aimed at dismantling the company, Huawei has become a symbol of China's technological self-reliance. Currently, Huawei is focusing on building a Chinese-style AI infrastructure with its Ascend chips and CloudMatrix system.
While there remains a significant gap in individual chip performance compared to Nvidia's H200 or the next-generation Blackwell system, particularly in memory bandwidth, power efficiency, and software ecosystems, China has chosen a different path. Even if it lags in individual chip competition, it aims to maximize overall system performance by clustering hundreds or thousands of chips together.
China's strategy can be summed up as “quantity over quality.” Even if power efficiency is somewhat lower and more space is required, it is pushing forward with massive capital and state support. Importantly, China is not solely targeting conversational AI like ChatGPT.
China aims to integrate AI across its entire manufacturing sector, a concept referred to as manufacturing AX (AI Transformation). This includes connecting AI to automotive factories, robotics, logistics, ports, power grids, smart cities, and the defense industry. This is not merely industrial innovation; it represents the construction of a new industrial civilization.
Another concerning aspect is cost. Chinese AI models have begun to establish a much lower cost structure than their American counterparts. By combining power, land, data, labor, and state support, China is rapidly reducing the costs of AI token generation. This ultimately means that numerous Chinese startups and manufacturers can adopt AI much more quickly.
Conversely, the US remains the world's leading AI nation. Top AI companies, including Nvidia, OpenAI, Google, Microsoft, Amazon, and Meta, are concentrated in the US. The country maintains a dominant advantage in advanced semiconductor design, AI models, cloud computing, operating systems, and global platforms.
However, the US faces a dilemma: losing access to the Chinese market. For companies like Nvidia, China represents a market that is hard to forgo. As the US government tightens export regulations, China accelerates its push for self-sufficiency, creating a situation where American companies may find it increasingly difficult to re-enter the Chinese market over time.
This scenario is what Jensen Huang fears most. While American GPUs are currently the best in the world, if China builds its own ecosystem, American products could lose their foothold in the Chinese market in a few years. In other words, US regulations could paradoxically promote China's self-reliance.
Amid these developments, the world is increasingly moving toward technological decoupling. In the past, during the era of globalization, US technology, Korean memory, Taiwanese foundries, and Chinese assembly factories were interconnected in a single supply chain. However, the likelihood of a split into US and Chinese blocs is growing.
Interestingly, ancient classics have long warned of the dangers of excessive power clashes. The Dao De Jing states, “What is too strong does not last long,” and Sun Tzu's Art of War advises, “The best victory is the one that is achieved without fighting.” Today’s US-China technological hegemony competition is likely to culminate in the question of how to create a coexistence order rather than a complete victory for either side.
The challenge lies with South Korea. South Korea is a world leader in memory semiconductors, with Samsung Electronics and SK Hynix effectively dominating the HBM market, which is crucial in the AI era. However, unlike the US, South Korea has not mastered platforms and AI models, nor does it possess the vast domestic market and state-led industrial policies like China.
Recently, slogans like “AI G3” and “AI Big 3” have emerged in South Korea. However, realistically speaking, becoming the 'third AI hegemon' after the US and China is quite challenging. The US has the world’s best platforms and capital, while China has a massive domestic market and a state mobilization system. South Korea does not have the scale to compete head-on with either country.
So, what should South Korea do?
First, South Korea must become a key player in AI infrastructure. The country’s real strengths lie in memory and manufacturing. In the AI era, a memory-centric structure may become more important than simple GPUs. Particularly in the inference AI era, power efficiency and resolving memory bottlenecks will be crucial. South Korea should lead the development of next-generation memory and packaging, as well as low-power AI semiconductor structures beyond HBM.
Second, it should adopt manufacturing AX as a national strategy. South Korea has a world-class manufacturing base in automobiles, shipbuilding, steel, semiconductors, batteries, and biotechnology. The ability to integrate AI with manufacturing may even give South Korea an advantage over the US. Rather than becoming an “AI platform nation,” South Korea should aim to be an “AI manufacturing innovation nation.”
Third, South Korea must maintain a strategic balance between the US and China. The South Korean economy is linked to both US technology and the Chinese market. Choosing one side could destabilize the entire industrial base. Therefore, South Korea needs a complex strategy that cooperates with the US on technological security while also maintaining connections with China in terms of market and industry.
Fourth, the direction of a South Korean sovereign AI should be realistically redefined. Competing with the US and China in the race for super-large general models is unlikely to be feasible. Instead, it is far more important to create world-class competitiveness in vertical AI optimized for specific industries such as manufacturing, healthcare, finance, defense, robotics, and logistics.
The Beijing summit between the US and China may have seemed quiet on the surface, but beneath it, a massive new Cold War surrounding AI and semiconductors has already begun. The US seeks to contain China, while China aims for self-reliance. In this context, South Korea is entering a period of its most challenging strategic choices in history.
The Thucydides Trap is no longer about warships and missiles; it is now about GPUs, HBM, data centers, power grids, AI models, and manufacturing AX. The next decade is likely to be a full-scale battle over who defines the standards of future industrial civilization.
Will South Korea remain a mere supplier of components in this turbulent environment, or will it leap forward as a key nation in the manufacturing revolution of the AI era? The time for choice has begun.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.