As aspects of AI, from training to inference depend upon the ability to store and move massive volumes of data efficiently. Training, inference, checkpointing, data preparation, model outputs and retrieval-augmented generation, RAG, all require digital storage that support modern AI demands for speed and scale. Let’s look at some recent AI-related announcements by Backblaze and CoreWeave, Panmnesia and Meta, as well as Cloudera and Vast.
Cloud storage provider, Backblaze recently signed an agreement with CoreWeave. The $335M agreement will have Backblaze providing cost-efficient storage capacity to support portions of CoreWeave’s managed storage infrastructure, helping optimize placement of data across performance tiers while preserving high-performance storage resources for the demands of AI workloads. Backblaze supplies HHD-based storage tiers in CoreWeave AI Object Storage.
In addition to HDD-based solutions from Backblaze, CXL memory advances for AI workloads are starting to bear fruit.
Panmnesia recently reported on results using its CXL controller and CXL switch chips, including a real-world deployment by Meta at the 2026 ISCA. The figure below shows CXL memory expansion for a CPU local memory.
According to the release, Panmnesia found that its next-stage CXL controller and PBR, Port-Based Routing, switch—both design-optimized and built in real silicon—keep memory-access latency low while significantly extending memory expansion to span dozens of servers or more.
Meanwhile, Meta reported that using CXL to expand memory in production services cut the number of servers needed for distributed AI inference by up to 25% and shortened the average response time of its distributed caches by about 29%, delivering gains in both cost and performance.
Cloudera and Vast Data announced a partnership to deliver AI data anywhere. This is a scalable production environment where data is continuously ingested, refined, governed and delivered to AI models for training and inference. It is available across on-premises environments and public clouds so that organizations can deploy AI services wherever performance, compliance and cost requirements are best met. This solution combines Cloudera’s containerized data services with the Vast AI Operating System.
According to the announcement, Cloudera’s lakehouse architecture provides portable, containerized data services, including data engineering, streaming, analytics, machine learning, and AI, all deployable consistently across hybrid and multi-cloud environments.
The Vast AI Operating System is based upon the Nvidia AI Data Platform reference design and unifies high-performance storage, database and global namespace capabilities for large-scale AI, analytics and mission-critical data workloads. Vast’s OS delivers exabyte-scale data infrastructure, integrating vector database services with Nvidia cuVS for GPU-accelerated vector indexing and search as well as high-performance storage optimized for modern GPU clusters and AI workloads.
The end result, according to the announcement, is that Vast’s AI OS leverages the NVIDIA AI Data Platform reference design, with Nvidia-accelerated computing, transforms latent enterprise data into activated AI-ready data, while Cloudera delivers data engineering, analytics, governance and AI services on top.
AI storage announcements by Backblaze, CoreWeave, Panmnesia and Meta, as well as Cloudera and Vast provide data where it is needed in the AI workflow.