DDN unveiled AI400X3M storage appliance and KV-cache solution optimized for Nvidia Vera Rubin platform AI inference workloads.

DDN has launched the AI400X3M storage appliance and a new KV cache solution in response to anticipated storage demands from Nvidia's Vera Rubin platform. The AI400X3M features significantly faster throughput than prior generations, designed to handle the intensive I/O requirements of large-scale AI inference workloads.

The KV cache solution integrates with Nvidia's middleware stack to optimize inference performance, addressing a critical bottleneck in serving generative AI models at scale. This pairing positions DDN to capitalize on the expected adoption surge as Vera Rubin—Nvidia's next-generation inference GPU—reaches customer deployments.

For the AI buildout, these products address a key infrastructure gap: the storage layer has traditionally lagged behind compute acceleration. By optimizing both raw throughput and the specialized caching mechanisms that inference workloads require, DDN is removing a potential constraint on how quickly inference clusters can scale. This is particularly relevant as data-center operators build out dedicated inference infrastructure to handle the computational load of deployed models.