US Semiconductor Tariffs Are a Trap for America Too
Washington's push to tax imported chips and tie tariff exemptions to US production is forcing a reckoning: American AI builders can't source HBM domestically fast enough, and the tariffs will hit their wallets first. But the real pressure on Seoul is years away—and it could redraw the global chip map.
Washington Wants Chips Made in America. It Can’t Have Them Yet.
The Trump administration is circling a new tariff strategy aimed squarely at South Korea’s semiconductor industry, and the timing lands uncomfortably close to a reality both sides feel but haven’t said aloud: there simply aren’t enough chips right now to go around.
According to PolitiKo, US Commerce Secretary Howard Lutnick is favoring a system where tariff-free import quotas would be set as a percentage of each company’s US-based production commitments. A maker that builds more fabs in America gets a bigger slice of untaxed imports. The framework first appeared in January’s deal with Taiwan, which lowered the reciprocal tariff on Taiwanese chips from 25% to 15% in exchange for TSMC pledging to build US capacity worth 2.5 times its current output—with an even higher multiplier once those fabs come online.
South Korea received most-favored-nation treatment in November, but only as a verbal promise, not a written agreement. That gap matters.
The Immediate Problem: Nobody Has Chips to Tax
The short-term economics work against Washington. Global memory supply—DRAM, NAND, and especially HBM—remains in acute shortage. Hyper-scaler capital expenditure on memory is projected to surge from roughly 8% of total AI server build costs in 2023–24 to around 30% this year, according to SemiAnalysis. Prices are climbing. Lead times are long.
Micron, the only US-headquartered memory maker, produces just a fraction of its DRAM domestically right now. The company has publicly stated it aims to manufacture 40% of its DRAM in the United States by 2035. That is not a timeline that helps a White House looking for leverage today.
A tariff on imported memory chips would land primarily on American buyers—NVIDIA, Amazon, Google, Microsoft—stacking cost on top of cost at a moment when building AI data centers is already one of the most capital-intensive pursuits in corporate history. That is what analysts are calling self-inflicted damage, and it is not hard to see why.
The Longer Game: A Trap for Seoul
Here is where the pressure actually builds.
Samsung and SK Hynix have confirmed plans to invest roughly 800 trillion won between them on expanded memory capacity in South Korea. That scale of investment proves two things to Washington: first, that Korean firms have the financial muscle to absorb US tariff-linked quotas; second, that they are dependent on continued access to the American market to make those expansions worthwhile.
Once the formula hardens—US production equals tariff-free quota—the calculus shifts. Korean chipmakers will face a choice that did not exist before: produce in America to protect market access, or stay put and risk being priced out by tariffs on their exports.
Samsung is already advancing a foundry in Taylor, Texas. SK Hynix has begun construction on an advanced packaging facility in Indiana for HBM. Neither operation is designed to replace domestic capacity; both are supplemental. But if the US framework tightens, supplemental may no longer be optional.
The Taiwan precedent is instructive. TSMC is building a fab in Arizona, yes—but it is also maintaining its leading-edge capacity in Taiwan. The US deal bought geopolitical insurance without forcing a full migration. For Korean companies, the equivalent would likely demand far more capital, far more land, and far more operational risk in an environment where talent pipelines for advanced memory fabrication are thin outside of East Asia.
Who Wins, Who Loses
Short answer: nobody in the next two years. The memory shortage is too severe for tariffs to bite meaningfully. America’s own AI ambitions run on HBM built in South Korea, and there is no alternative source at scale.
Longer answer: the companies that get ahead of the quota game win. If the US settles on a production-linked quota model similar to Taiwan’s, SK Hynix and Samsung will compete not just on technology and price, but on how much American capacity they can promise. The firm that commits most aggressively to US investment will enjoy the largest tariff-free quota and the greatest leverage in Washington. The firm that hesitates will face shrinking margins on its American sales—or worse, exclusion.
There is also a secondary channel of pain. Korean home appliance makers—Samsung Electronics and LG Electronics—sell smartphones, PCs, and servers in the United States. If chipflation pushes component costs higher and tariffs layer on top, those product categories become less price-competitive in the American market. This is indirect pressure on Korea, but it is real.
What Comes Next
The administration has not finalized details. PolitiKo reports that the plan remains under review and subject to significant revision. That ambiguity itself is a form of leverage—it keeps Korean firms guessing and potentially overcommitting to US expansion before the rules are locked in.
What is clear is that Washington has chosen a path that treats semiconductor access as a bargaining chip rather than a strategic dependency. The memory shortage will ease eventually. As it does, the tariff framework will matter more. By then, the question for Seoul will not be whether to invest in America, but how much of its future capacity it is willing to locate there to preserve market access.
That is a decision no Korean company wants to make under tariff threat. But it is exactly the kind of decision this policy is designed to force.