DDN Preps for AI Wave With Speedy New Appliance, KV Cache Solution
AIwire, Wednesday, July 1st, 2026
DDN launches the AI400X3M appliance and a distributed KV cache acceleration solution to ready storage for AI inference.
DDN expanded its AI and HPC data platform portfolio with the new AI400X3M appliance and a distributed KV cache acceleration architecture integrated with NVIDIA Dynamo.
The appliance raises write throughput about 35 percent to roughly 190 Gbps, boosts random reads 4x to 8 million IOPS, and supports 30PB per rack, including hybrid NVMe and disk arrays to ease NAND supply constraints.
The KV cache solution stores and retrieves model context from DDN's platform instead of local GPU memory, which DDN claims delivers up to 55x faster KV cache loading.
The moves target growing AI infrastructure demands as platforms like NVIDIA Vera Rubin arrive.