Kioxia's Sold-Out 2027 Can't Close the Gap Between Forecast and Expectation

Kioxia has now put a number on the awkward gap between a boom and what investors will accept as proof of one. The memory maker's production for 2027 is fully committed, according to the company — yet its guidance for the December quarter landed below consensus, and the stock fell 14% on the day.

The shortfall itself is modest in absolute terms. Kioxia projects operating profit of roughly ¥2 trillion for October to December 2026, against a Bloomberg-reported consensus of ¥2.35 trillion. What unsettles the market is less the profit figure than the distance between the company's own outlook and the bar analysts had set.

A forecast that barely moves the needle

The guidance also offers little sense of acceleration. For the July–September quarter of fiscal 2026, Kioxia reported revenue of ¥2.48 trillion and operating profit of ¥1.9947 trillion. Its revenue target for the following quarter is ¥2.5 trillion — an increase of just 0.8% quarter on quarter, with capacity shipments expected to stay roughly flat.

That leaves investors with an unusual combination: a business running at extraordinary levels, and a forecast that suggests the next step up will be small. A strong quarter, in other words, does not automatically produce a strong surprise.

The same pattern shows up in pricing. Average NAND selling prices climbed 70% in the April–June quarter, then rose by roughly 25% — the mid-twenties in percentage terms — in July–September. Reports on the company's outlook point to broadly stable prices in the current quarter. Growth is decelerating, but prices are not falling; the two are different things, even if a market with high expectations can treat the first as though it were the second.

What the AI boom doesn't fix

Daiwa attributes the disappointing guidance in part to weak smartphone and PC demand and to cautious pricing assumptions. Kioxia itself cites pulled-forward orders, the stronger yen and longer delivery times as drags, while expecting a reacceleration in the fourth fiscal quarter.

That is a useful corrective to the simplest version of the AI-storage story. Heavy demand from data centers does not mean every end market is expanding at once, and currency effects or shipment timing do not disappear simply because customers are locking in capacity years ahead. Near-term revenue depends as much on when product ships and at what price as on whether buyers want it.

The longer view looks sturdier. Long-term contracts already cover more than half of expected shipments for 2028, and Kioxia is expanding at Yokkaichi and Kitakami, with the K3 fab slated for 2029. Customers are securing supply over extended periods while the company prepares additional output — a structural shift that reaches well beyond the next quarterly print.

A first dividend, and a question mark

Alongside the guidance, Kioxia declared an interim dividend of ¥90 per share — its first payout since listing, according to Reuters. The distribution answers a different question from the forecast: it concerns how shareholders share in the company's success, not how the operating business will develop.

Capacity expansion and Chinese competition have both been flagged as potential headwinds. Neither amounts to an imminent glut, but together they explain why long-term demand commitments cannot fully settle the question of future pricing.

The read-across is narrower than the headline move suggests. Kioxia's outlook does not refute AI-driven memory demand; it refutes the assumption that scarcity produces uninterrupted earnings acceleration. The next test for investors is whether contracted long-term demand translates back into faster near-term growth — and that, rather than any single session's decline, is what will determine whether the stock's 266% gain so far this year holds up.