Two customs ledgers, three products, one uncomfortable answer.
The story is finished before you hear it. China is giving up on batteries. The world is queuing for Korean cells. A 72-trillion-won empire is cracking. American manufacturers are shifting lines to Chungju because the Pentagon bans Chinese batteries outright from October 2027, and Korea is the safe pair of hands. Every part of that is sourced, quoted and plausible. It is also the kind of story nobody checks, because checking is dull and the conclusion already feels earned.
If the decoupling worked anywhere, it worked here. So who is actually selling batteries to America?
You are right to expect a Korean answer. Tariffs are in place. The Inflation Reduction Act tilts the field. The Pentagon has published a hard cut-off. There is no friendlier ground anywhere in world trade for the thesis that supply chains have moved. We went and read the American customs ledger for May 2026, expecting to confirm it.
China is first. Not narrowly — by twelve points. And Korea, the country described as the destination of the great migration, is third, behind Japan.
Decoupling did not happen to a country
Run the same query one product over. In integrated circuits, HS 8542, China is eighth at 2.5%, behind Taiwan’s 40.1%. Here the split is real. Move one more, to transformers and static converters — the unglamorous iron that grids are made of — and China is third at 10.3%. Same country, same month, same tariff regime. First, eighth, third. The split did not happen to a nation. It happened to a product.
That is the American view. We keep the Japanese one, and it reads as a mirror. Japan’s battery exports to Greater China fell from 16.7% in 2021 to 4.5% this year — a quarter of what they were. Read alone, that is a collapse. Held against the American ledger, where Japan is the second-largest battery supplier at 18.7%, it is something else entirely. It was not a loss. It was a move.
Japan’s chips went the other way, to 66.4% — above the 56.0–62.6 band its own completed years have held since 2021. A move needs a destination. American battery imports had room: three suppliers sharing sixty-five percent, and Japan walked in. American chip imports had none. Taiwan alone takes 40.1%; Japan is seventh at 4.5%. No door, so the share stayed where it was and grew.
Decoupling was sold as risk reduction: spread the eggs, lose fewer of them. In the one product where it worked, American chip imports now run 77.0% through five suppliers, with Taiwan alone at 40.1%, while Japan’s chip exports run 66.4% to Greater China. Both sides are standing in a narrower room than before.
The split did not distribute the risk. It sorted it into two piles and made each pile deeper.
What this does not say: customs records where goods went first, not who finally used them. Shipments through Hong Kong are re-exported; Korean memory sent to Taiwan is packaged and leaves again as an American accelerator. A destination share is a routing fact, not a consumption fact. We measure shares and values, not margins or contracts — a ledger tells you who paid, not who won. And 2026 is a partial year: if the completed figure lands below 62.6, the band holds, the break disappears, and this reading fails. We will record that too.
Data: U.S. Census Bureau International Trade (public domain, CC0) and Japan Ministry of Finance customs statistics. Figures are May 2026 for US imports and full-year shares for Japan. Findings may be cited with attribution to Asia Pulse.