Kioxia Q2 FY26: ¥2,480.0B at an 80% Margin, and a Flat Guide That Isn't Flat

The setup

Kioxia's flat December guide is a currency effect, not a price peak. In dollars, management is guiding revenue up 3.4%. Our call is that December-quarter revenue lands at or above the ¥2,500.0B (~US$15.53B) guide, and we are wrong if it comes in below ¥2,390.0B, the level the September quarter was guided to in July, because that would mean NAND contract prices turned down.

The September quarter itself was not close. Revenue was ¥2,480.0B (~US$15.40B), up 40.3% on the quarter and 453.2% on the year. Operating profit was ¥1,994.7B (~US$12.39B), an 80.4% operating margin, and profit attributable to owners was ¥1,355.7B (~US$8.42B). Gross margin reached 82.8%. All of it came from a company that reports one segment, memory, and that spent ¥84.1B (~US$0.52B) on property, plant and equipment in the quarter, less than it booked in depreciation and amortization.

Put the headline against its own history before reading it as a quarter. Over the last cycle Kioxia's best quarter was Q2 FY24, at roughly ¥480.9B, derived as FY2024 revenue of ¥1,706.46B less the other three quarters. Six quarters ago, in Q1 FY25, revenue was ¥342.8B and gross margin was 20.8%. September's revenue is about five times the prior peak and seven times the trough. This is not a good quarter inside a normal range. It is a different business size, produced almost entirely by price, at a factory that has not grown.

The July dive on this site, Kioxia Q1 FY26, carried management's September guide of about ¥2.39tn of revenue and ¥1.89tn of operating profit, and the reading that "price and profit first" would hold, with new fabs running well below full output. Grade it now: revenue beat the guide by 3.8% and operating profit by 5.5%, and capex for the half came in at ¥135.3B against ¥153.5B of D&A. The call held. The utilization figure we carried in July is not in the Tanshin, so that half of it stays unscored.

Kioxia released the results after the Tokyo close on October 30, so the shares have not reacted yet. They closed that day at ¥13,470, down 25.6% from the end of September and 39.5% from the end of June on a split-adjusted basis. The market has already priced a peak. This page argues the filing does not show one.

The numbers

Every figure above is from Kioxia's Tanshin for the six months to September 30, 2026, filed October 30, and the Tanshins for the prior quarters. The Tanshin is unaudited and unreviewed, as the filing itself says. Quarterly cash flows and quarterly gross profit are derived: the six-month figure less the first quarter. Kioxia's fiscal year ends March 31, so Q2 FY26 is July to September 2026. Dollar conversions on this page use ¥161 to the dollar, the September-quarter average the company reports, and are approximate.

Two lines need context before the argument starts. Operating profit beat non-GAAP operating profit in the quarter because stock-based remuneration was a ¥6.4B reversal, a gain, after a ¥19.4B charge in the first quarter. And the gap between operating profit and profit is mostly tax: ¥613.0B (~US$3.81B), a 31.1% rate on profit before tax of ¥1,968.7B. Almost none of that tax has been paid yet, which matters in the cash section.