Kioxia Has Lost Half Its Value and Nobody Is Calling It a Korea Story

Kioxia's market value has halved from its peak, a fall of about 181 billion dollars, and most of it happened in three weeks. The stock dropped 16 percent on the 17th of July, 18.33 percent on the 28th, and another 13.85 percent on the 29th to 38,380 yen. Korea got the headlines this month because Korea tripped circuit breakers. Japan's leg of the same unwind is arguably larger in dollar terms.

SoftBank has been the other pressure point. It fell 9 percent on the 17th, 4.43 percent on the 28th to 5,095 yen, and 6.95 percent on the 29th. That is a holding company whose value is essentially a bet on the AI capital cycle, and it is being repriced as that cycle gets questioned rather than for anything specific to its own results. Its stake in Arm and its exposure to the OpenAI buildout are the two lines everyone is marking.

The index numbers hide the concentration. The Nikkei 225 fell 3.95 percent on the 28th, 1.49 percent on the 29th to 61,434.19, then rose 0.7 percent on the 30th to 61,867.43. That is a bad few days, not a crash. Underneath it two of the market's largest AI-linked names lost double digits repeatedly, which is what a rotation looks like when the index does not tell you.

The trigger was not Japanese. It came from a Reuters report on the 27th that a Shanghai firm has begun producing immersion DUV lithography scanners, and from ChangXin Memory listing the same day at a 466 percent gain to become China's most valuable listed company. Both are claims about 2028 supply. Kioxia sells NAND. Its multiple was built on scarcity lasting.

What makes it awkward is that the scarcity is still there. Samsung reported NAND average selling prices up in the high sixties quarter on quarter and told its call the memory shortage worsens through 2027 and into 2028. SK Hynix had NAND up 50 to 55 percent after 70 the quarter before. The physical market did not change. The story about how long it lasts did.

So Japan has a valuation problem rather than an earnings problem, which is harder to fix because there is no quarter you can report your way out of. Kioxia's market value halving on a competitor's IPO and a lithography rumour is a market saying it no longer trusts the duration of the boom. That is a different thing from doubting the boom, and it is worth keeping the two apart.