atehpengaday
09-10 08:00

While Kioxia ruling out deeper SK Hynix ties and resisting further NAND price hikes introduces a temporary pricing headwind, it does not mean the structural memory supercycle is over. SK Hynix and Samsung’s finished inventories dropping below 10 days—coupled with HBM4 production cannibalizing standard DRAM wafer capacity—highlights unprecedented physical supply constraints across AI datacenters. Kioxia’s strategy to protect long-term market share may cap short-term spot price spikes in NAND, but severe structural tightness in high-bandwidth memory (HBM) and enterprise DRAM continues to underpin SK Hynix's structural earnings power.

Two Sessions Undid Friday's Rally: Memory Supercycle Over?
Memory split: SK Hynix +4.83%, Micron −1.61%, SanDisk −0.12% — the same names that rallied together on Friday, SanDisk +12%, Micron +6%. The tightness is real: the shortage is spreading from HBM into DRAM and NAND, and Korean brokers put Samsung's and SK Hynix's inventories below ten days. The crack: Kioxia denied merger talks with SK Hynix and signalled it would cool price rises — the opposite of the tight-supply story. The counterparty to higher prices is not just the customer but rivals who want to keep that customer. Follow SK Hynix on inventory, or wait for Kioxia's stance to hit quotes?
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

  • LisaEffie
    09-10 08:32
    LisaEffie
    HBM and enterprise DRAM tightness still looks structural to me. HBM4 starts are eating standard DRAM wafer slots, so even flat AI demand keeps baseline supply tighter than people think
Leave a comment
1