Chips are now 47% of Korea's exports. What happens when prices turn?

In the first ten days of August 2026, South Korea shipped $21.29 billion of goods. Semiconductors were $9.95 billion of that — 46.8% of everything the country sold abroad, according to Korea Customs Service provisional figures released on 11 August. A year earlier the same share was 26.7%.

That is the highest chip share ever recorded for a ten-day period, and it is the number the wire copy led with. The more useful question is what the other 53% has been doing, because the answer changes how you should read every Korean macro headline for the rest of the year.

The boom is one product line

Set the ten-day figure aside for a moment and look at complete periods, where the working-day noise washes out.

Shares calculated from the reported totals. Sources: MOTIR annual and monthly releases; Korea Customs Service for the August period.

Two things fall out of that table. First, the share has more than doubled in nineteen months — from roughly a fifth of exports to something approaching half. Second, semiconductor exports in the first half of 2026 alone, at $192.4 billion, already exceeded the previous full-year record of $173.4 billion set across all of 2025.

The other half of the picture is quieter than the headlines suggest. The Ministry of Trade, Industry and Resources reports that non-semiconductor exports rose 16% in the first half of 2026 and 26% in July. Those are good numbers. They are not extraordinary numbers, and they are the ones that describe the shipyards, the chemical plants, the steel mills and the car factories where most Korean manufacturing employment actually sits.

Most of the growth is a price, not a shipment

This is the part that consistently goes missing in coverage of Korean trade data, and it is not a matter of interpretation — the ministry says so in its own release.

MOTIR attributes the semiconductor surge to AI server investment and “higher memory contract prices.” Contract prices are the fixed prices at which memory makers sell to large buyers, reset periodically. When they rise, the dollar value of an unchanged number of shipped chips rises with them.

The same release makes the value-versus-volume gap explicit elsewhere. In July 2026, petroleum product exports rose 34.1% by value while volumes fell 8.8%. Petrochemical exports rose 10.3% by value while volumes fell 9.1%. In both cases Korea shipped less and earned more, because prices moved.

Dollar export totals, in other words, are partly a price index. That is always true; it is unusually true in 2026, in the one category that now accounts for nearly half the total.

What a price reversal would actually cost

Nobody publishes a credible forecast of memory contract prices, and this piece will not invent one. But the exposure can be sized with arithmetic on published figures.

July 2026 exports were $98.89 billion, of which chips were $41.01 billion. Hold every other line steady and vary only the semiconductor line:

Author’s calculation holding non-semiconductor exports at their July 2026 level of $57.9bn. Illustrative, not a forecast. Base figures: MOTIR, July 2026.

Note what the last row implies. Even a full round-trip in chip export values — back to where they stood a year ago — would leave Korea’s monthly exports around $72.6 billion, which is above any month before 2026. The non-chip economy has genuinely grown. What would vanish is the growth rate, the trade surplus of the size Korea is currently running, and a large part of the earnings that have driven Samsung Electronics and SK hynix share prices. For who supplies what in that market, see our breakdown of who actually supplies the AI memory behind the 2026 boom.

The surplus is the line to watch. Korea’s cumulative trade surplus for January–July 2026 was $168.0 billion, against $33.9 billion in the same period of 2025 — an improvement of $134.1 billion. A country running that surplus has a great deal of policy room. A country whose surplus is one product line’s contract price has considerably less of it than the number suggests.

The destination map is re-concentrating

Between 2024 and 2025 Korea’s export mix diversified: ASEAN’s share rose from 16.7% to 17.3%, exports to Latin America and the CIS grew fastest, and reliance on both China and the United States fell. MOTIR made a point of this in its 2025 annual release.

Chips have partly reversed it. In July 2026 exports to China rose 96.2% to $21.68 billion and exports to the United States rose 68.7% to $17.43 billion — the two markets that buy the AI infrastructure the memory goes into. ASEAN, at $18.80 billion, set an all-time monthly record, but much of that is also semiconductors moving to assembly and test sites.

So the same force driving the export record is quietly rebuilding the concentration that Korean trade policy spent two years trying to reduce — and doing it at a moment when Minister JK Kim’s own July statement flags “new tariff measures under Section 301 of the U.S. Trade Act” as a live risk to the outlook.

How to read the ten-day numbers without being fooled

Korea Customs Service publishes provisional figures for the first ten and first twenty days of each month, and they are genuinely useful — nowhere else gets you trade data that fast. They are also extremely noisy, and the August release is a good illustration.

In the 1–10 August window, car exports fell 80.9% to $182 million. That is not a collapse in the Korean car industry; car exports rose 7.0% in July and shipments are lumpy, because a single vehicle carrier sailing on the 11th rather than the 9th moves the whole line. The same window showed exports to the United States down 0.2% while July’s monthly figure was up 68.7%.

Applied here: the 46.8% chip share is real, and it is the highest reading on record, but the defensible statement is that chips are running somewhere in the low-to-mid 40s as a share of Korean exports, and rising. July’s 41.5% is the more reliable anchor.

What this means for anyone with exposure to Korea

If you hold Korean equities, the index is now close to a leveraged position on memory contract prices. That is not a new observation about the KOSPI, but the export data shows how far the concentration has gone at the national-accounts level rather than just the index level — and Korean equities have already demonstrated in 2026 how quickly that can unwind, as our account of the 2026 KOSPI crash sets out.

If you are watching the won, the record surplus has not translated into currency strength, and the reasons are structural rather than cyclical; the mechanics are in our piece on why the won stays weak while exports boom.

If you work in Korea outside the chip supply chain, the macro headlines are describing a boom you are unlikely to feel directly. Non-semiconductor export growth of 16% in the first half is solid, ordinary performance, and the sectors carrying most employment — cars, machinery, petrochemicals, steel — grew in the low single digits or shrank.

The honest summary of Korea’s 2026 trade data is that it records two economies. One is having the best year in the country’s industrial history on the back of a global capital-spending cycle it does not control. The other is having a decent, unremarkable year. The published totals average them into a single triumphant number, and the gap between those two realities is the risk that Korean policymakers, and anyone with money in Korean assets, is actually carrying.

Figures current as of 14 August 2026. Export and trade totals are from Ministry of Trade, Industry and Resources monthly and annual releases; the 1–10 August period is Korea Customs Service provisional data and is subject to revision. Share calculations are the author’s arithmetic on those published totals.