en.Wedoany.com Reported - On June 4, Japanese storage company Kioxia announced that it will participate in Interop Tokyo 2026 from June 10 to 12 at Makuhari Messe, showcasing flash memory and SSD technologies for generative AI, data centers, and mobile applications. Exhibits include the 245.76TB enterprise SSD KIOXIA LC9 series, Super High IOPS SSDs for GPU and AI applications, the software technology KIOXIA AiSAQ designed to improve the accuracy of generative AI responses, and the 3D flash memory BiCS FLASH technology.
The focus of Kioxia's exhibition is to further push flash memory and SSDs from traditional data storage devices into the AI system performance chain. As the scale of generative AI applications expands, data centers not only require higher computing power but also higher capacity, lower latency, and more stable data read capabilities. Large model training, inference, retrieval-augmented generation, vector databases, and enterprise knowledge base applications all place higher demands on storage systems. The 245.76TB enterprise SSD can increase data density within limited rack space, while the Super High IOPS SSD addresses the need for high-speed data scheduling in GPU-intensive tasks. For AI infrastructure operators, the capacity, throughput, latency, and thermal performance of storage devices are now jointly determining system efficiency alongside GPUs, network interconnects, and memory configurations.
The theme of Interop Tokyo 2026 is "The Next Chapter of AI and the Internet." Kioxia will showcase its latest products and technologies at its booth and conduct live demonstrations.
This exhibition also involves the ShowNet interconnection verification network. Kioxia will provide the latest SSDs supporting liquid cooling, including the KIOXIA CM9 series and KIOXIA CD9P series, to participate in interconnection verification of network and server cooling solutions in the era of high heat generation. As AI server power consumption continues to rise, data center cooling methods are rapidly evolving from traditional air cooling to liquid cooling, water cooling, and hybrid cooling. Although SSDs are not the highest power-consuming components, they still generate thermal pressure in high-density storage, continuous read/write, and compact server environments. Including water-cooled SSDs in ShowNet testing indicates that storage devices are also entering the thermal management system of AI infrastructure, no longer just serving as backend capacity components.
KIOXIA AiSAQ reflects the trend of storage companies extending into the software layer. The quality of generative AI responses is related to model capabilities, as well as external knowledge access, retrieval speed, data organization, and context provision. By combining SSDs with software technology to improve the data access and response quality of AI applications, Kioxia signifies that storage vendors are shifting from "selling hardware capacity" to "supporting AI data flow." If such technologies can be more closely integrated with enterprise knowledge bases, private data platforms, and AI inference systems in the future, the role of SSDs in AI applications will expand from underlying storage to data availability and inference efficiency.
For Japan's storage industry, Interop Tokyo remains an important window for showcasing network, server, data center, and internet infrastructure technologies. Kioxia's concentrated presentation of large-capacity SSDs, high-IOPS SSDs, AI software technologies, and water-cooled storage solutions reflects that flash memory suppliers are repositioning product value around AI-era internet infrastructure. Future competition will focus not only on single NAND capacity and interface speed but also on system-level cooling, GPU collaboration, data retrieval efficiency, and AI workload adaptability.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
