Prices doubled, shipments fell: can Kioxia sustain the ¥1.5tn+ in owner earnings its stock implies?

In the January–March 2026 quarter, Kioxia’s revenue hit ¥1,002.9bn, roughly double the prior quarter, and non-GAAP operating income jumped about fourfold to ¥599.1bn. Yet the earnings presentation shows that bit shipments in the same quarter fell about 10% quarter on quarter. Revenue doubled on lower volume because average selling prices more than doubled.

For the full year ended March 2026, revenue rose 37.0% to ¥2,337.6bn and net income attributable to owners of the parent climbed 103.6% to ¥554.5bn, both records. The market, meanwhile, values the company at more than ¥30tn. At a 5% required return, that price implies more than ¥1.5tn of owner earnings every year, about 2.6 times the ¥583.6bn the company actually generated last fiscal year. Can a record built on a price spike be sustained at 2.6 times its own level for years? That gap is what this note measures.

A cost race inside the NAND oligopoly

Kioxia is a pure-play NAND flash maker that traces its roots to Toshiba’s memory business. It mass-produces 3D NAND, branded BiCS FLASH, at large fabs in Yokkaichi and Kitakami, and sells it as SSDs for data centers and PCs and as embedded memory for smartphones and other devices. Memory is its only reportable segment. By application in the year to March 2026, SSD & storage brought in ¥1,362.6bn (58.3%), smart devices ¥760.0bn (32.5%), and other ¥215.0bn (9.2%).

The profit formula is simple: bits shipped × average selling price − cost per gigabyte. The only durable way to protect margins is to spend heavily on capex, move to higher layer counts, and keep pushing cost per gigabyte down. Against Samsung, SK hynix (Solidigm), Micron, and Sandisk, Kioxia’s claim to a wide moat rests on world-class manufacturing scale and stacking technology: eighth-generation BiCS FLASH is in volume production and the tenth generation is on the way.

Its AI-server lineup has also deepened. It includes the CM9 enterprise SSD, the 245.76TB LC9 high-capacity QLC drive, a Super High IOPS SSD aimed at NVIDIA’s next-generation storage architecture, and SSDs for KV cache. The common goal is to raise GPU utilization and get designed into customers’ systems. R&D spending was ¥141.1bn.

The weak spots are just as clear. The product is close to a commodity, and prices follow supply and demand. The annual securities report itself cites the NAND price collapse of late 2022 through 2023 as a risk. There is almost no recurring revenue, and manufacturing depends on joint-venture fabs shared with Sandisk. The company also admits that the replacement of nearline HDDs is moving more slowly than it first expected.

Revenue grew 37% on price, not volume

The 2025 and 2026 columns are the fiscal years ended March 2025 and March 2026. Revenue rose from ¥1,706.46bn to ¥2,337.63bn, net income from ¥272.32bn to ¥554.49bn, and the net profit margin from 16.0% to 23.7%. The earnings release attributes the growth to a sharp rise in average selling prices, plus higher bit shipments, driven by strong demand from data center customers, mainly for generative AI.

The margin jump is a fixed-cost business running in reverse. Depreciation and amortization barely moved, from ¥312.3bn to ¥312.8bn, so most of the price increase fell straight to profit. Non-GAAP operating income reached ¥876.2bn, a 37.5% margin. When fab costs are roughly fixed, rising prices lift profit sharply; falling prices shrink it just as fast. This looks less like a structural step-up in earning power and more like strong market conditions absorbing a heavy fixed-cost base.

The momentum carried into the new fiscal year. According to the first-quarter earnings release published on July 31, 2026, April–June revenue was ¥1,767.1bn, non-GAAP operating income was ¥1,326.2bn (a 75.0% margin), and quarterly net income attributable to owners of the parent was ¥842.2bn. One quarter out-earned the entire previous fiscal year. Net income still came in 3.1% below the company’s ¥869.0bn guidance, because of a ¥36.6bn provision for litigation losses and ¥19.4bn of share-based compensation. Management’s second-quarter outlook calls for revenue of ¥2,390bn and non-GAAP operating income of ¥1,900bn.

How far ¥583.6bn sits from the ¥1.5tn a ¥30tn+ market cap implies

(* Simulation assumptions: estimated share price 55,780 yen, required return 5%)

The estimated share price is on a pre-split basis. Kioxia’s 1-for-3 stock split took effect on October 1, 2026; on a post-split basis the price is about 18,593 yen.

Owner earnings (net income + depreciation − capex) rose 62.6%, from ¥359.022bn to ¥583.616bn. That is slower than the 103.6% rise in net income because depreciation stayed flat while capex (tangible plus intangible) increased from ¥225.6bn to ¥283.7bn.

Owner earnings value at a 5% required return is ¥11,672.32bn, or about ¥11.7tn. The simulated market cap at 55,780 yen times shares outstanding is about ¥30.5tn (the live market cap was about ¥31.8tn as of October 2, 2026), so owner earnings value covers only about 38% of it. Turned around, 5% of a ¥30.5tn market cap is about ¥1.53tn a year. The stock is priced as if Kioxia can generate roughly 2.6 times its latest annual owner earnings on a permanent basis.

The company also received ¥56.4bn in government subsidies in the year to March 2026, which effectively lightened its capex burden. Even if those subsidies are netted against capex, owner earnings come to about ¥640bn, which barely narrows the gap to ¥1.5tn+. The first-quarter run rate suggests the number is not out of reach. But that only holds if prices stay at their newly doubled level for several years.

On a conservative basis, net cash is still around zero

Net cash here uses a conservative definition: (current assets − inventories) + investment securities × 70% − liabilities. At the end of March 2026 it was -¥932.314bn, an improvement of about ¥751.6bn from -¥1,683.912bn a year earlier, but still negative. The net current asset ratio was -3.054%.

The quality of that improvement matters. Trade receivables rose by about ¥422.0bn, from ¥238.6bn to ¥660.6bn, which explains more than half of the gain. The fourth-quarter revenue surge went straight into working capital. On the financing side, Kioxia repaid ¥616.4bn of long-term borrowings and redeemed ¥323.0bn of non-convertible preferred shares, while refinancing with ¥535.6bn of new loans and ¥326.7bn of U.S. dollar bonds.

Applying the same definition to the first-quarter balance sheet in the earnings release gives net cash of roughly +¥73.5bn. It has turned positive, but against revenue of this size it is essentially zero. Trade receivables rose further to ¥1,151.6bn at the end of June. The equity ratio attributable to owners of the parent reached 50.8%, and the senior loan is on track to be repaid in full, so the balance-sheet rebuild is moving fast. Still, if prices fall, this improvement can unwind as receivables are collected at lower levels. That is the assumption to hold.

What bites if prices turn

The biggest risk is the NAND cycle itself. If the price spike of the fourth quarter and first quarter turns out to be the peak, the heavy fixed-cost base means profit will fall as fast as it rose. The company expects demand to exceed supply in calendar 2027, but that is a company forecast, not a fact.

The second risk is renewed capex expansion. Gross capex for the year to March 2027 is planned at ¥450bn, about 1.6 times the ¥281.1bn spent on property, plant, and equipment the year before. One company adding capacity to meet demand is reasonable; the whole industry adding capacity at once could break the tight-supply assumption. Higher capex also eats into owner earnings until depreciation catches up.

On the demand side, management itself notes that PC shipments are falling as bill-of-materials costs rise, and that low-end smartphone volumes are shrinking. Kioxia depends heavily on data centers and smartphones, and it is exposed to currency moves (the average rate was 150 yen per dollar in the year to March 2026) and U.S. tariff policy. The nature of the ¥36.6bn litigation provision booked in the first quarter, and whether more could follow, could not be confirmed from the available materials. Inventory days stood at 103 in the fourth quarter, inflated by strategic DRAM stock.

The share price swings as hard as the earnings. On a split-adjusted basis, Nikkei data shows the stock rising from around 3,648 yen on January 6, 2026 to around 37,566 yen on June 22, then closing at 19,300 yen on October 2.

There is real upside potential too. Products such as KV-cache SSDs and Super High IOPS SSDs for AI inference move Kioxia a step beyond being a pure bit supplier. Data center and enterprise SSDs set records in both volume and revenue in the fourth quarter. If growth shows up in volume, not just price, the assessment would change.

Wait: confirm it with volume and cash, not price

Kioxia holds a place in the NAND oligopoly with world-class scale and can compete head-on for AI storage demand. For now, though, the stance here is wait. There are two reasons. First, most of the record profit came from a spike in selling prices, while shipments actually fell. Second, a market cap of about ¥30.5tn implies about ¥1.53tn a year in owner earnings, and price durability has not yet been demonstrated well enough to treat 2.6 times last year’s ¥583.6bn as sustainable earning power. Conservatively measured net cash is also around zero, so it offers no real margin of safety on the downside.

Bullish monitoring triggers

• Bit shipments of data center and enterprise SSDs grow quarter on quarter, so revenue growth is explained by volume as well as price. Check the application-level revenue and shipment commentary in the presentation for the second quarter of the year to March 2027 (results scheduled for October 30, 2026).
• Conservative net cash widens from roughly +¥73.5bn at the end of the first quarter, and trade receivables do not grow faster than revenue. Check the quarterly balance sheet.

Warning / pass triggers

• Company guidance signals a quarter-on-quarter decline in average selling prices, or inventory days rise well above the 103 days of the fourth quarter.
• The ¥450bn capex plan for the year to March 2027 is raised significantly, or the company softens its calendar-2027 outlook that demand will exceed supply.

Record profit plus a swing to positive net cash is not a reason to relax. Rather than annualizing a 75% first-quarter operating margin to justify the share price, the plan is to wait until volume growth and receivable collection show up in the numbers.

Want to simulate the "fair value" of Kioxia Holdings Corporation?

Interested in digging deeper into the financial data for Kioxia Holdings Corporation featured in this article?

At okuriru.com, you can run simulations for Kioxia Holdings Corporation based on Buffett-style "owner earnings." Enter your required rate of return (hurdle rate) to instantly calculate a theoretical stock price.

👉 Calculate Kioxia Holdings Corporation's theoretical stock price

Resolve the question "Is this stock fairly priced?" using your own criteria. CSV downloads of financial data are available too!