FMS 2026, aka the Future of Memory and Storage, is taking place this week from August 4 through to 6 in Santa Clara, California. As the inventor of flash memory and a leader in storage technology, KIOXIA will be on hand to showcase the next generation of PCIe 6.0 NVMe SSD storage built for cutting-edge AI infrastructure. KIOXIA will also be conducting a keynote presentation, panel discussions, and various technical sessions and demonstrations focused on performance, scalability, and efficiency.
According to the company, these three pillars define modern AI workloads, from enterprise through to cloud and high-performance computing. And by leveraging the right storage solution, companies and organizations can remove performance bottlenecks, improve efficiency, and scale more effectively. One of the new products that will be highlighted at the show is the company's new KIOXIA GP1 Series PCIe 6.0 NVMe SSDs, which deliver an impressive 10 million random read IOPS and are optimized for GPU direct access.
The company will also be showcasing its new KIOXIA CM10 Series PCIe 6.0 enterprise SSDs and KIOXIA NX1 Series SSDs, which support direct liquid cooling. The KIOXIA CM10 Series is also built for modern AI workloads, with 92% higher sequential read and around 85% higher random read performance compared to the previous-generation CM9, and they're the first SSDs to use the latest BiCS FLASH generation 10 TLC flash memory technology with 332 stacked layers.
"AI is rapidly evolving across every industry, and memory is at the center of that transformation," said Scott Nelson, Executive Vice President and Chief Marketing Officer for KIOXIA America, Inc. "At KIOXIA, we're enabling the next generation of AI with flash memory and storage technologies that deliver the performance, density, and efficiency organizations need. From the data center to the edge, our innovations are a game changer, helping power the AI applications transforming how businesses operate and compete."
Outside of new hardware, the company's keynote presentation, titled "Redefining Data Center Storage for the AI Era with Flash Memory and SSDs," will kick off on Tuesday, August 4 at 11:10 am, focusing on AI's transformation from model training to inference and how that has dramatically impacted requirements for both memory and storage. KIOXIA will also host a special session with NVIDIA, titled "Balancing Memory and Storage to Operate AI Factories and Agents at Scale," that will take on a panel/Q&A format featuring NVIDIA and other industry leaders.
Frequently Asked Questions
Open a question for an answer from TweakTown's coverage of this news, or ask your own below.
What performance improvements does the CM10 Series offer versus the CM9 Series for sequential and random reads?
Which KIOXIA product families showcased at the event are specifically promoted for GPU-direct access or near-GPU workloads?
What flash memory technology and layer count does the CM10 Series use?
What demonstrations or sessions at KIOXIA's booth relate to scaling AI inference and vector databases?
Have a question that isn't listed here? Ask below, and TweakBot will answer it.
KIOXIA will also be there on the show floor with a massive two-level booth (#307), broken down into the following exhibitions.
- Liquid Cooling SSDs for AI - featuring KIOXIA NX1 Series Data Center NVMe SSDs
- Context Memory Caching for Maximizing AI Inferencing - featuring KIOXIA CM9 Series Enterprise NVMe SSDs
- PCIe 6.0 NVMe Enterprise SSDs - featuring KIOXIA CM10 Series Enterprise NVMe SSDs
- KIOXIA GP1 Series Super High IOPS SSD - 10 million IOPS in 512-byte random read performance for near GPU Workloads
- AI RAG: Improving Scalability for Vector Databases - featuring KIOXIA AiSAQ open source software
- High Capacity QLC Storage for Fast Retrieval of Large Datasets - featuring 245.76 TB KIOXIA LC9 Series NVMe SSDs
- New UFS 5.0 - AI Ready, High Performance Memory Solutions
- CXL Memory Module with KIOXIA XL-FLASH - expanding the memory hierarchy with XL-FLASH
- BiCS FLASH generation 10 Flash Memory - advanced scaling with CMOS directly Bonded to Array (CBA) technology, featuring 332 layers with enhanced lateral shrink
- High Capacity Package/Low Latency QLC Flash Memory - featuring a 32-die stack BGA package for AI applications
