South Korea's Impossible Position

Photo: Unsplash

Alliance Politics

South Korea's Impossible Position

Seoul must simultaneously protect its security alliance with Washington and its economic lifeline with Beijing — the semiconductor war has made that balancing act untenable.
south-koreasemiconductorsus-chinaalliance-politicssamsung

In July 2023, the South Korean government announced that Samsung and SK Hynix would be allowed to continue operating their existing chip manufacturing facilities in China without seeking annual license renewals, despite US export controls that would normally require them. The US granted an indefinite waiver. Both governments described this as a pragmatic accommodation. What it actually was is a precise illustration of how the semiconductor war forces every party except the two principals into positions where every choice is simultaneously necessary and insufficient.

South Korea’s semiconductor industry is not incidental to its economy. Samsung Electronics accounts for roughly 20 percent of South Korean exports. SK Hynix is the world’s second-largest DRAM producer. Together, the two companies collectively define a significant portion of the global memory chip supply — the type of chip that goes into every smartphone, server, and increasingly, every AI accelerator that requires high-bandwidth memory. China is both a customer for these chips and a manufacturing location for both companies, which have invested tens of billions of dollars in Chinese fab capacity over the past two decades.

Washington’s demand, applied with escalating intensity from 2022 onward, was that South Korea restrict its chip companies’ sales and technology transfers to Chinese customers and Chinese manufacturing operations. The strategic logic was clear from the American perspective: the export controls only work as a system if the allies hold the line. Restricting Nvidia sales while allowing Samsung to keep supplying DRAM to Chinese AI chip makers is like cutting off the oxygen while leaving the fuel line intact.

Seoul’s position was equally clear, even if less loudly stated. China is South Korea’s largest trading partner. The entire structure of Korean economic growth over the past thirty years has been built on deep integration with the Chinese manufacturing ecosystem. Samsung’s Chinese fabs represent capital investments that cannot be written off without material damage to the company’s balance sheet and the Korean government’s domestic political support. And the security calculus is complicated by the geography: North Korea shares a border with South Korea, and China is the primary external actor capable of moderating North Korean behavior. Antagonizing Beijing is not an abstract foreign policy choice for Seoul. It has specific and immediate implications for the threat environment that Koreans live with.

The result has been a policy of partial compliance with strategic ambiguity. South Korea restricted some technology transfers, complied with the letter of export control requirements when compliance was unambiguous, and negotiated carve-outs and waivers that preserved Korean company operations in China to the maximum extent that American patience would allow. This is not a betrayal of the alliance. It is the behavior of a country making rational decisions under genuinely irreconcilable pressures.

The high-bandwidth memory issue is where the pressure has been most acute. HBM — the specialized memory chips that AI accelerators require to feed data to their processing cores at the rates that make large model training possible — is produced in meaningful volumes by exactly two companies: Samsung and SK Hynix. The US wants those chips restricted from Chinese AI hardware developers. China is one of the largest markets for those chips. Korean companies are caught in the middle of a supplier-customer relationship that was profitable and is now politically fraught.

SK Hynix, which has been somewhat further from Chinese operations than Samsung, moved to comply more fully with US restrictions in 2024 and 2025. Its Chinese operations were wound down more aggressively. Samsung, with deeper Chinese manufacturing investments and more complex relationships with Chinese device makers, moved more slowly. By 2026, the divergence between the two companies’ postures had become a visible tension in Korean industrial policy discussions — with Samsung arguing for more accommodation and SK Hynix, having already paid the costs of tighter compliance, arguing for a clearer alliance position.

The 2026 tightening of US export controls resolved some of this ambiguity by making the rules more explicit, which is a mixed outcome for companies that had been operating in the grey zone. Explicit rules are easier to comply with and harder to negotiate around. Korean chips are now more firmly restricted from the Chinese AI hardware market than they were two years ago. The revenue impact is real, and it is absorbed differently by companies with different exposure profiles.

What makes South Korea’s situation politically distinctive is the domestic politics of alliance loyalty. American and European commentary tends to assume that being a formal US ally means aligning with US policy positions on matters the US considers strategic priorities. South Korean public opinion is more complicated. The alliance with the US is popular and supported across most of the political spectrum. But the economic relationship with China is also popular, in the sense that Korean prosperity depends on it, and Korean voters understand this even when they cannot articulate it in policy terms.

The Korean political system has a tradition of what analysts call “strategic ambiguity” — maintaining alliance relationships while preserving room for independent economic diplomacy. This is not unique to Korea; Japan has practiced a version of it for decades. But Korea’s position is more exposed because its economy is less diversified than Japan’s, its security situation is more immediately threatened, and the geographic proximity to China means that both economic and security consequences of any alignment choice are faster to arrive.

The presidential election of 2027 will test whether the semiconductor-related compliance costs are large enough to have shifted Korean domestic politics. The conservative government that approved the more aggressive HBM restrictions did so over significant business community pushback. If a progressive government wins in 2027 and seeks to renegotiate the terms of semiconductor export compliance, it will discover — as every government in this position discovers — that the leverage Washington holds is not primarily about chip policy. It is about the totality of the security relationship, including the American troop presence on the peninsula and the extended nuclear deterrence guarantee that makes South Korea’s defense posture viable.

The broader pattern that South Korea exemplifies is one that most analyses of the US-China technology war underweight: the degree to which the competition’s outcomes will be determined not by bilateral US-China decisions but by the choices of the countries caught in between.

Taiwan, the Netherlands, South Korea, Japan — these countries collectively control capabilities that neither the US nor China can replicate on short timelines. They are not passive actors in the semiconductor competition. They are active decision-makers with their own interests, their own domestic politics, and their own calculations about which alignment maximizes long-term security and prosperity. The US has treated them, at various points, as assets to be secured rather than partners to be persuaded. That approach works until it doesn’t, and it tends to stop working at precisely the moment when the pressure is highest.

South Korea’s semiconductor companies will continue to navigate this environment with the pragmatism that Korean industrial culture has historically displayed under adversity. Samsung survived the 1997 Asian financial crisis, the 2008 global recession, and the OLED patent wars with Apple — not by resolving the fundamental contradictions in its competitive position, but by managing them with enough tactical skill to keep building. The semiconductor cold war is a harder problem than any of those. But the instinct is the same: adapt, survive, preserve optionality, and wait for the political environment to clarify.

That is a rational strategy. It is also, in the meantime, an exhausting one.