Kioxia Won't Chase Market Share: Is the Storage Boom Still On?

This news brings the risk of price hikes reaching end-users to the table, but the storage market itself hasn't cooled down because of it. It's suitable for those interested in AI infrastructure costs and storage cycles to examine closely, but not appropriate as a short-term tip for immediately bottom-fishing or selling off.

I followed my usual process for investment/financing news. On September 9, I saw this message from IT Home. The headline was strong: Kioxia CEO Yuuta Ota denied deepening cooperation with SK Hynix, stating that price increases were already high enough and he wouldn't chase market share for the sake of it. I didn't judge immediately; instead, I threw the keywords into Perplexity, which I'd started using two days prior. Source cards appeared, including Bloomberg summaries, Sina Finance, Wall Street Insight, and Gate News. The first round summary was quick, saying Kioxia refuted building deeper ties with SK Hynix while promising to control memory price spikes to avoid harming long-term AI demand.

Here, a term needs explanation. NAND is flash memory, common in phones, SSDs, and servers; HBM is High Bandwidth Memory, more critical in AI servers. Storage chip price hikes directly impact data center and end-user costs. Strong AI demand doesn't mean customers are willing to accept unlimited prices.

Next, I asked DeepSeek to help pull a timeline. I've used DeepSeek for about a month; it handles Chinese news summaries decently. I input key dates: August 29, SK Hynix CEO said no fixed plan for large indirect stakes in Kioxia; August 31, SK Hynix denied reports of joint factory construction in Japan; September 2, Chey Tae-won said joint production with Kioxia was one option, also studying building a factory in Japan; September 9, Yuuta Ota denied deepening cooperation. It arranged things clearly, distinguishing between research, options, denials, and plans by intensity. The snag was one version where "studying" was written as "considering advancing," making the tone heavier. I reviewed the summaries line by line to correct it.

Later, comparing several reports, I found the controversy centered on how deep the relationship was. SK Group and Kioxia already have capital and historical links; industry rumors long suggested joint production, Japanese factories, and combined NAND shares exceeding Samsung. Reports mentioned that if cooperation landed, their combined NAND market share might surpass Samsung. This judgment is stimulating because it implies global storage landscape changes. On the other hand, SK Hynix emphasized only conducting relevant research, mentioning no joint ventures, power, or water resources. Kioxia's CEO directly denied deepening cooperation, with a clearer attitude.

When reading such news, I usually ask first: who is leaking info, and who is putting out fires? Chey Tae-won's words sound like strategic options, easily read by the market as prelude to M&A or alliance. Yuuta Ota's words look more like price management and expectation management. The material notes memory chip makers expanding capacity for AI orders, with demand growth pushing price hikes to double or even triple digits. If true, pressure transmits from manufacturers to buyers.

I also asked ChatGPT, which gave a bearish view, only reminding me to break judgments into trackable indicators.

My testing shows the most valuable part is Kioxia bringing the price issue to the forefront. Over the past year, Kioxia's stock rose rapidly, with reports claiming cumulative gains over twentyfold. With sentiment so high, the CEO talking about not chasing share or hyping cooperation is essentially cooling things down.

My judgment leans conservative. Suitable for those caring about AI servers, storage cycles, and Japanese semiconductor capital moves; unsuitable for short-term traders just wanting to hear about giants joining forces and immediate explosions.

What really needs watching in this news is storage prices starting to make AI clients frown.