Kioxia Posted a Record Quarter. The Market Sold It Anyway.

Kioxia Holdings did almost everything a semiconductor company is supposed to do. It posted the highest quarterly operating profit in its history, produced an operating margin that its own industry rarely sees, and announced its first dividend since listing. And on Monday morning in Tokyo, its shares fell as much as 15 percent. The gap between what the company reported on Friday evening and what the market did with it on Monday is the story of the week in Asian technology, and it is not a story about a bad quarter. It is a story about a very good quarter that arrived with a cautious shape attached, from a finance chief who would not promise the good times would keep rolling.

Kioxia Posted a Record Quarter. The Market Sold It Anyway.

Tokyo, Japan — November 2, 2026 — The Japanese memory maker's US depositary receipts had already dropped 17 percent in response to its outlook before Tokyo opened, according to Bloomberg. When trading began, Kioxia fell as much as 15 percent from Friday's close of 13,470 yen to 11,450 yen, a level confirmed by Nikkei. Japanese markets are closed on Tuesday for a public holiday, so Monday's intraday prices are the last word investors will get for a day.

The Record Nobody Wanted

Kioxia makes NAND flash memory, the storage chip inside solid-state drives, phones and data-centre racks. For the three months to the end of September it reported revenue of 2.48 trillion yen and an operating profit of 1.9947 trillion yen. That profit is a company record, and more than 22 times the year-earlier level according to Futunn. The half-year is stranger still. Revenue for April through September came to 4.2472 trillion yen, up 436.8 percent from 791.1 billion yen a year earlier, and operating profit reached 3.2647 trillion yen, 25 times the 130.8 billion yen posted in the same period last year. The operating margin was 76.9 percent, which the Seoul Economic Daily called "an unusual level of profitability for a chipmaker." Those are the numbers of a company selling something scarce into a market that cannot get enough of it. They were released at 18:45 Japan time on Friday, after the close, and by Monday morning they had been comprehensively rejected.

What Kioxia Actually Told the Market

The rejection was not about the past quarter. It was about the next one. Kioxia guided October through December operating profit to about 2 trillion yen, roughly 12.7 billion US dollars, against a consensus of 2.35 trillion yen. Revenue was guided to 2.5 trillion yen against a market expectation of 2.9 trillion yen, and net profit to 1.36 trillion yen against an expectation of 1.62 trillion yen, according to the Economic Daily News in Taiwan. The company also forecast operating profit of 5.2597 trillion yen for the nine months from April through December, 19 times the year-earlier level, against a market expectation of 5.5428 trillion yen, Yonhap reported via Financial News. Every one of those figures is enormous in absolute terms. Every one of them sits below what analysts had pencilled in. That is the whole problem in three lines, and none of it is a fall in profit. It is a flattening.

Seventy Percent to Twenty

The mechanism behind the miss is the price of NAND flash, and the direction of travel is what matters. Average selling prices rose about 20 percent quarter on quarter in July through September, which Bloomberg put in the mid-20 percent range. That sounds like a boom until you compare it with the preceding quarter, when prices rose 70 percent. The rate of increase has collapsed while the absolute price stays high. Akira Menju, Kioxia's head of finance, said it is unclear whether prices can keep rising in fiscal 2027 given how much selling prices have already climbed. That is not a forecast of a downturn. It is an admission that the second derivative has turned, and in a market that had priced the first derivative as permanent, the second derivative is the only thing anyone trades on.

The Question UBS Asked

That caution ran straight into the company's own supply commentary, and one analyst at the briefing noticed the collision. Kenji Yasui, head of research at UBS Securities Japan, asked how Kioxia could describe supply and demand as extremely tight in the next fiscal year while calling its price outlook uncertain at the same time. The answer did not satisfy the room, or at least it did not satisfy the market. The tension is real rather than rhetorical: Kioxia said it has secured buyers for all of the volume corresponding to its production capacity for next year, and has signed long-term supply contracts with hyperscalers for well over half of its expected shipments in 2028. Those are the words of a company with pricing power. Yet the guidance implies that pricing power will not translate into much sequential growth this quarter. Investors were left to reconcile the two, and they chose the cautious half.

A First Dividend, Deliberately Small

Kioxia also declared its first dividend since listing, an interim payout of 90 yen per share totalling about 48 billion yen. The company had initially planned no payout at all, so this is a genuine change of policy rather than a routine confirmation. The market treated it as insufficient, and SMBC Nikko Securities, in a note cited by the Economic Daily News, said the dividend total is small. Against a business generating operating profit at trillion-yen scale every three months, 48 billion yen reads as a gesture rather than a return of capital. The credit side of the story is stronger: S&P upgraded Kioxia to BBB on 13 October, per a company news release. But a ratings upgrade and a modest first dividend do not offset a below-consensus guide in a stock that had spent the year priced for perfection.

Two Brokerages, Two Readings

The sell-side did not speak with one voice. SMBC Nikko calculated that revenue is guided to rise only about 1 percent quarter on quarter, with currency moves cutting roughly 2.5 percent from the top line, shipments measured by storage capacity flat and average selling price up only slightly. It called the pricing assumption conservative relative to the market and to peers, warned that more smartphone product shipments could dilute the overall average price, and noted that changed customer demand could push some data-centre and enterprise products into January-March shipments, improving volume and pricing early next year. Citigroup was blunter, calling the results disappointing and the October-December outlook below even the recently darkened market expectation. But it found a reassurance where SMBC Nikko found arithmetic: Kioxia appears to expect January-March shipments to rise nearly 10 percent quarter on quarter, unusual against normal seasonality. Citigroup also said a 1 trillion yen buyback this quarter or soon would be difficult given heavier second-half capital spending, cash accumulation and corporate income tax, and that a share-price reversal needs either data-centre investment worries to fade or a buyback to arrive.

The One That Went Up

While Kioxia was falling, Tokyo Electron was rising. The equipment maker reported its own April-September results after the close on Friday, and the numbers were strong across the board: revenue of 1,667,455 million yen, up 41.3 percent year on year, and operating profit of 506,776 million yen, up 67.2 percent. It then disclosed full-year guidance to March 2027 for the first time, and the guidance was the point. It projects operating profit of 1,169,000 million yen, up 87.1 percent, on revenue of 3,700,000 million yen, up 51.4 percent. It set an annual dividend of 523 yen per share, which Kabutan calculated is an effective 52.1 percent increase after the one-for-five split carried out in September. Traders Web recorded the divergence at the opening bell: Tokyo Electron up sharply, Kioxia down sharply. The obvious inference, and it is an inference rather than anything the companies said, is that a maker of the machines that build memory fabs is being paid for capacity while a maker of the memory itself is being repriced for price.

Sandisk Went First

Kioxia was not the first pure-NAND maker to be punished for a good quarter last week. Sandisk, which runs Japanese fabs in a joint venture with Kioxia and develops the same technology alongside it, reported a day earlier. September-quarter revenue came to 11.031 billion US dollars, up 23 percent on the prior quarter and 378 percent from a year earlier, against about 10.7 billion expected. Adjusted earnings per share were 49.51 dollars before a split, against about 47.15 expected, and adjusted gross margin was 85.1 percent. Data-centre revenue reached 4.8 billion dollars, up 61 percent sequentially, while consumer revenue fell 29 percent to 397 million. December-quarter guidance of 12.2 to 12.6 billion dollars landed at or above consensus, and the company announced an eight-for-one forward split effective 19 November. CEO David Goeckeler cited "robust and growing demand for our products driving outpaced growth in our data center business." Sandisk told investors the demand under negotiation in long-term deals is larger than everything it can make through the end of the decade, Yahoo Finance reported. The stock still closed down 5.32 percent at 1,330.05 dollars and fell about 9 percent in Friday morning trading.

And Then New York Happened

Friday in New York gave the market a second reason to sell memory. Intel plunged nearly 30 percent, its steepest single-day drop in more than four decades, according to Trading Economics, and the Philadelphia Semiconductor Index fell 7.2 percent. Western Digital dropped 8 percent, Seagate 7 percent and Micron 6 percent, Yahoo Finance reported. Reuters attributed the US losses to late earnings from Amazon.com and Intel that showed massive cash burns. There was a credit story in the background too: Blackstone told investors that the AI developer Anthropic may not receive an investment-grade credit rating immediately after listing, the Financial Times reported via Yonhap. And there was a political calendar. US midterm elections are on Tuesday, with results expected during Asian trading hours on Wednesday, and polls suggest Trump's Republican Party may lose control of at least one chamber of Congress, per Bloomberg. Shoji Hirakawa, chief global strategist at Tokai Tokyo Intelligence Lab, said the disappointing Kioxia results will likely stoke fears that Japanese tech earnings will also lag expectations.

Where the Numbers Actually Stood

Kioxia was down 12.36 percent at 11,805 yen in a TradingKey snapshot of the Asian session, having touched 11,450 yen earlier. The broader market was softer without being broken. The Nikkei 225 opened down 0.50 percent at 62,972.02 and was later about 0.3 percent lower near the 63,000 level, against Friday's close of 63,290.11, when it had fallen 2.1 percent. Topix was about 0.8 percent lower early. In Seoul the KOSPI fell 1.10 percent to 5,667.49, with Samsung Electronics down 3 percent at 218,500 won and SK Hynix down 1.4 percent at 1,399,000 won. Ibiden fell 6 percent and Screen Holdings 5.8 percent. Banks were hit as well, Mitsubishi UFJ down 2.3 percent, Sumitomo Mitsui 2.6 percent and Mizuho 2.5 percent. SoftBank Group was down 3.19 percent at 4,338 yen after falling 6 percent on Friday. The Nikkei had fallen almost 13 percent from its June record in the period through 30 October, with Kioxia and capacitor maker Taiyo Yuden each having lost more than half their value, per Bloomberg. That is the backdrop against which a record quarter was sold.

What Nobody Can Say Yet

Several things are being asserted this week that the evidence does not support. Kioxia has made no statement explaining the share move and has not commented on the selloff, and there is no confirmed closing price for Kioxia, the Nikkei, Topix or KOSPI for Monday, because trading is still under way. What none of it establishes is that the memory cycle has peaked or that prices will fall. The guidance shows slower price growth and a below-consensus quarter, nothing more. Even the forecasters do not agree among themselves. TrendForce expects fourth-quarter NAND contract prices to rise 15 to 20 percent, yet has warned that supply could exceed demand in the second half of 2027. Kioxia expects demand to exceed supply in 2027. S&P said the shortage could last until at least 2028. Bloomberg Intelligence expects January-March operating profit may resume growth and sees 2027 NAND demand by storage capacity growing in the low-20s percent, with a richer server mix lifting the price of each unit of capacity. Four respected houses, four different pictures of the same market, and a share price that moved before any of them were proved right.

By Jessica Ali, Staff Writer

This article was produced with AI-assisted research and editorial support. Sources: Kioxia Holdings investor relations, Nikkei, Bloomberg, Reuters, Seoul Economic Daily, Economic Daily News, Yonhap via Financial News, Futunn, Traders Web, Kabutan, Trading Economics, TradingKey, Yahoo Finance, the Financial Times via Yonhap, SMBC Nikko Securities, Citigroup, UBS Securities Japan, TrendForce, S&P Global Ratings and Bloomberg Intelligence. Photographs: Seoul Economic Daily, Yahoo Finance and TradingKey. Cover: Global 1 News editorial image.