- Kioxia describes its plans for a new SSD for AI workloads reaching 10 million iops
- IOPS measures how fast a storage device can handle small, random requests
- New Drive uses XL-FLASH, a type of SLC NAND and a new internal controller
Kioxia has revealed plans for a new SSD that stands could hit an impressive 10 million IOPS, a level of performance that has been square on the requirements of AI-run systems.
SSD uses XL-FLASH, a type of cell (SLC) NAND level) combined with a new internal controller.
A spokesman for Kioxia told TechPowerup“We take our ultra-fast XL-Flash memory chips that use cells at one level and pair them with a whole new controller … We are targeted at over 10 million IOPs and we plan to have trials ready in the second half of 2026.”
Difference between IOPS and GBPs
IOPS or input/output operations per Second measures how fast a storage device can handle small, random requests, especially important in AI and server applications, how quick access to small files is key.
This is different from GBPs that refer to the actual data transfer rate and is used to measure how quickly large files can be read or written.
A drive with high GBPs may excel in video editing or large file transfers, but for machine learning tasks where thousands of small data packages are read or written constantly means high IOPS more.
Kioxia’s approach to the next resurrection includes not only one -off projects, but a broader effort to accommodate cases with different uses. Its CM9 series, sampling for customers now, focuses on speed and reliability to match advanced GPUs used in AI, while the LC9 series provides massive 122TB capacities for large databases.
Behind these products is the 8th generation Bics Flash, which introduces CBA Tech to increase performance and efficiency.
Kioxia also prepares future flash memory generations using two methods. The first adds more layers to capacity, while the second mixes new CMOS design with older cell structures to keep investment costs in check.
