Koreas Backchannel Play to Keep Its Chip Crown

The real story behind a private dinner

President Lee Jae-myung did not invite Samsung Electronics Chairman Lee Jae-yong to a formal policy briefing. They met over food, quietly, on August 26. Two weeks earlier, the president had done the same with SK Group Chairman Choi Tae-won. Hyundai Motor Group CEO Cha Eon-hyon is next in line.

This is not a social calendar. It is a sequence — a deliberate, backchannel campaign to align Korea’s three largest chip-dependent conglomerates with White House expectations before Washington writes the next invoice.

The choice of format is itself a signal. Formal state visits and public announcements strip away nuance. A private dinner leaves room for concessions that neither side could easily afford to make on camera. Lee Jae-myung, still consolidating authority in his first full year as president, is using these occasions to extract commitments that the MKE (Ministry of Economy and Finance) and KOSTAT would struggle to deliver through bureaucratic channels alone. The chaebol chairmen, for their part, gain something in return: direct access to the president without the interference of competing ministries or the press.

Why Washington is still calling

American officials, led by Commerce Secretary Howard Lutnick, have been pressuring Seoul for an additional investment commitment beyond what was already agreed under the original US-Korea semiconductor partnership. The demand: more new fab construction on American soil.

The argument from Washington is straightforward. Korea’s Samsung and SK Hynix dominate the memory chip market — roughly 70 percent of global DRAM and a commanding share of NAND flash — yet their biggest US capacity expansions are either planned or under construction in Texas and Arizona. America wants them to accelerate.

But the pressure runs deeper than the visible fab pipeline. Inside the Commerce Department and the National Security Council, there is a growing conviction that the original deal — announced in late 2023 — gave Korea too many exits. Samsung committed to $170 billion in US investment, but much of that is spread across multiple phases spanning a decade. SK Hynix pledged $60 billion, similarly phased. Washington now wants tranches unlocked sooner, and it is signaling that further export-control exemptions for Korean fabs operating in China could be conditioned on American expansion.

Korea’s counter-argument is equally straightforward: its own survival depends on staying competitive in a market where HBM (high-bandwidth memory) is becoming the single most valuable component in AI chip production. Every billion dollars shipped to Texas is a billion not spent on advanced nodes in Pyeongtaek and Giheung. Korean engineers working on the next generation of HBM4 are based almost entirely in South Korea. Move the capital, and you risk moving the expertise — or worse, creating a two-tier company where the American fabs receive marginal capacity while the Korean sites hold the strategic R&D.

The three mega-projects

Seoul has framed its response around what it calls three mega-projects: semiconductors, physical AI, and a third initiative whose details remain intentionally vague. The architecture is clear even if the specifics are not — the government is attempting to create an industrial strategy that satisfies both American investors and Korean competitiveness.

The 3,000 trillion won investment program announced in late June was supposed to be the centerpiece. But the private dinners suggest the public numbers were not enough to lock in the chaebols. Samsung’s board, in particular, has been cautious about overseas expansion that dilutes R&D at home. Sources close to the company indicate that Lee Jae-yong has repeatedly pressed for clearer guarantees on US market access for Korean-made chips before approving further American capital expenditure. SK Hynix faces similar internal debate, particularly around how much of its HBM output should be reserved for American customers versus the broader global market.

The president appears to be working directly around institutional resistance. By meeting chaebol leaders one-on-one, he is creating political capital that the MKE cannot easily ignore when the time comes to allocate subsidies, tax breaks, and regulatory fast-tracks.

Second-order effects: the ripple through the ecosystem

What makes this situation fragile is not just the relationship between Seoul and Washington. It is the downstream impact on Korea’s entire semiconductor ecosystem. Samsung and SK Hynix are not isolated giants — they anchor a network of hundreds of smaller suppliers, materials companies, and equipment integrators clustered around Pyeongtaek and Giheung. Move too much capacity abroad, and that cluster begins to atrophy. Companies like SEMCO, Hanmi Semichem, and Doosan Robotic Engineering rely on proximity to the major fabs. Their survival is tied to the Korean sites staying competitive, not just profitable.

There is also a labor dimension. Korea is already facing a severe shortage of semiconductor engineers. The country produced only about 1,200 new microengineering graduates last year, down from a peak of nearly 2,000 a decade ago. Building fabs in Texas does not solve that problem — it exacerbates it, because the homegrown talent pool has fewer domestic opportunities and some will look elsewhere. Japan and Taiwan are actively recruiting Korean-trained process engineers with signing bonuses that Korean companies struggle to match.

Who wins, who loses

If Seoul succeeds in splitting the difference — securing US concessions while keeping Korean fabs on Korean soil — the winners are the Korean economy and the global supply chain, which benefits from not having all its memory production relocated. The losers would be the American legislators who view semiconductor offshoring as a strategic vulnerability and want visible, tangible factory construction as proof of commitment.

If the talks collapse, Washington could escalate tariff threats or tighten export controls on equipment that Korea needs for its own advanced node development. That would hit Samsung and SK Hynix harder than any American company, because Korea has no meaningful alternative supplier for EUV lithography equipment. ASML is the sole provider, and the Netherlands follows Washington’s lead on export restrictions. A disruption there would be catastrophic for Korean production timelines.

But there is a less obvious loser in either scenario: the global AI ecosystem. Memory is the bottleneck for next-generation AI training and inference. If Korean investment is diverted or delayed by geopolitical friction, the cost is not borne in Seoul or Washington — it is borne by every company racing to build the next generation of AI models. That creates a counter-pressure from the technology sector that may yet shape the outcome.

The China question everyone ignores

There is a second layer to this dynamic that rarely surfaces in English-language reporting. China accounts for roughly half of South Korea’s chip exports. Beijing has been quietly tightening restrictions on advanced semiconductor equipment exports to China through third-country channels. Any American pressure on Korea to move production to the US carries implicit implications for how Seoul manages its relationship with the Chinese market — a relationship that Beijing treats as non-negotiable.

Samsung has already suspended some operations at its Xi’an facility. SK Hynix faces similar constraints. The question now is whether Washington expects Korea to fully decouple — and whether Seoul can afford to.

Full decoupling would mean closing or winding down Chinese operations entirely, a move that would cost Samsung an estimated $8 to $10 billion in annual revenue and eliminate thousands of high-skilled jobs in Gyeonggi Province that depend on the Chinese supply chain. No Korean government can make that calculation without severe political cost. Beijing has made this clear through quiet diplomatic channels and through the leverage it holds over rare earth exports, which are essential to chip manufacturing.

What happens next

The meeting with Hyundai Motor Group signals that the strategy extends beyond chips. Physical AI — the integration of artificial intelligence with robotics and automated systems — is the third mega-project, and Hyundai is the natural vehicle for that domain. This suggests Seoul is trying to build a broader industrial coalition, not just manage a single-sector dispute with Washington.

Hyundai brings something the chaebols do not: a foothold in the American auto market and a history of successful US investment through Hyundai Motor Manufacturing Alabama. The president may be counting on Hyundai to serve as a bridge — proof that Korean industrial expansion in America can coexist with Korean technological leadership at home.

What English-language readers should watch for is not the next dinner — those will remain private by design — but the announcement schedule. If Samsung or SK Hynix declares a major US investment within weeks, the bargaining position was weak. If instead Seoul announces expanded domestic capacity with US equipment partnerships that satisfy Washington without moving factories, the backchannel strategy succeeded.

There is also a timing question. The US midterm elections are two years away, but Washington does not wait for electoral cycles to apply pressure. Export controls are administrative tools — they can be tightened or loosened without congressional approval. That means Seoul has a narrow window to negotiate before the next round of restrictions lands.

The chips at the table were real. The strategy underneath them is what matters. And the measure of success will not be how much Korea invests in America — it will be whether Korea remains the place where the next generation of memory chips is actually designed.