South Korea's semiconductor exports could come in between US$50 billion and US$100 billion below earlier forecasts if the largest American technology companies slow their artificial intelligence capital spending next year, the Bank of Korea has said. The warning was reported by Cynthia Kim for Reuters in a despatch from Seoul carried by The Star.
The figures come from a written response by the central bank to the lawmaker Ahn Do-geol, rather than from a formal forecast round.
The bank set out what it called a pessimistic scenario. It assumes a sudden shortfall in exports accompanied by a 10 per cent decline in the unit prices of Korean semiconductors.
The exposure is unusual in scale. Chips make up roughly half of South Korea's exports, which is why a change in the purchasing plans of a handful of foreign companies registers as a national macroeconomic variable.
Under the scenario, next year's GDP growth would be about 0.4 percentage points lower than it would otherwise have been.
The current account would also be affected, with the surplus pushed well below the central bank's initial projection of US$430 billion for 2027. The bank did not publish a revised figure.
The bank set out the spending it is watching. Combined capital expenditure by five large American technology firms - Amazon, Google, Microsoft, Meta and Oracle - stood at US$40 billion last year, is estimated at US$80 billion this year and is projected at US$110 billion next year.
It also attached a caveat. The bank warned of considerable uncertainty over the trajectory of AI demand and over the broader memory chip business cycle, which has historically turned on its own schedule rather than on anyone's announcements.
The exercise is a conditional rather than a prediction, and the bank published a range rather than a single number.
That range, however, is now the most precisely specified object in the Korean outlook. It has a floor of fifty billion dollars and a ceiling of a hundred billion, which gives a spread of fifty billion and a midpoint of seventy-five. It has an assumption underneath it priced at ten per cent off a chip. Pessimism, in other words, has been quoted.
It has also been quoted one-sided. No equivalent figure has been published for the case in which American capital spending rises as projected, or rises faster. The optimistic scenario exists, but it has not been given a number, a spread or a midpoint, and cannot therefore be compared with anything.
On the published evidence the Korean economy's downside is bracketed to within fifty billion dollars and its upside is undefined. In any other market this would be described as a quote with no offer.
Our analysts note the further convenience. Should exports arrive within the range, the scenario will have been confirmed; should they arrive above it, the considerable uncertainty will have been confirmed instead. The memory cycle, which was named in the caveat, is available to account for either.
The Bank of Korea has published a number for everything that might go wrong next year and nothing for anything that might not, which makes pessimism the best-documented position in Seoul.