Kioxia plays its "negotiation trump card": Expansion plans require the signing of long-term contracts!

Author | Kozmon

      Editor | Hard AI

Recently, Kioxia and SanDisk jointly announced a 5 trillion yen expansion plan through fiscal year 2033. In its latest research note released on August 28th, UBS believes the key message is: Kioxia has the capacity to expand production but intends not to act immediately—this will be its ace bargaining chip with mega-scale cloud providers in NAND long-term agreement (LTA) negotiations.

The ongoing extension of AI model context length is rapidly driving up NAND demand for inference, with no signs of slowing down.

UBS estimates that the supply of NAND in 2027 will be tighter than in 2026. In negotiations for long-term agreements, mega-scale clients are likelier to make concessions, allowing Kioxia to secure favorable terms.

How will 5 trillion yen be spent?

Of the 5 trillion yen investment, Kioxia will contribute 60% and SanDisk 40%. 1.8 trillion yen is earmarked for the new K3 factory, with land preparation underway and production planned for fiscal year 2030. The target site for the remaining 3.2 trillion yen has not yet been disclosed.

The investment will be carried out in two stages. In fiscal years 2027-2029, the combined three-year total remains at the previously announced 1.4 trillion yen, with an annual average of roughly 450-500 billion yen, though some scheduled 2029 spending will move up to 2027 and 2028. From fiscal years 2030-2033, annual investment may surge to about 1.25 trillion yen.

UBS estimates that this investment will add monthly capacity of 200-220 thousand wafers for the 332-layer BiCS10 generation, and 180-200 thousand wafers for the 4xx-layer generation. Taking a monthly capacity of 350 thousand wafers at the end of fiscal year 2028 as the baseline, the completed investment could restore capacity to 550 thousand wafers, with minimal risk of market share decline. Kioxia has also slightly increased its bit growth forecast from about 20% to just above 20%.

Why is this the "ace" in negotiations?

Mega-scale customers naturally want Kioxia to expand production early to secure supply in long-term agreement talks. However, Kioxia’s latest announcement signals the opposite: expansion capability and plans are publicly disclosed, but the production timeline is deliberately slow.

In a tight supply environment, keeping capacity "on hold" shifts leverage to the supply side. UBS thus expects Kioxia to secure favorable terms in long-term agreements—mega-scale clients hoping to lock in future NAND capacity will have to give ground on price and contract duration.