WEKA

AI Infrastructure · Data Storage · Enterprise Tech

Launch

WEKA unveils NeuralMesh 6 and WEKApod 3, its first custom AI storage hardware

Extending GPU memory with cheap flash storage could reshape the economics of running AI agents at scale.

• WEKA launched NeuralMesh 6, a major software overhaul, alongside WEKApod 3, its first self-designed storage hardware line built specifically for AI inference and agentic workloads.
• WEKA calls itself an AI data and memory infrastructure company; its core idea is using cheap NAND flash to extend costly GPU memory via a system it calls Augmented Memory Grid.
• A single WEKApod 3 rack delivers 1.1 exabytes of effective capacity, 10.2 TB/s throughput and 210 million IOPS, making it the first single-rack system to break the exabyte barrier.
• On Oracle Cloud Infrastructure's H100 systems, NeuralMesh benchmarks show 10x higher token throughput, 10x more concurrent users served, and 7x more tokens per GPU versus a DRAM baseline; CoreWeave saw 4.2x more tokens per GPU.
• The launch places WEKA alongside Dell, NetApp, Pure Storage, VAST Data and DDN, all of whom have repositioned their storage lines toward AI infrastructure in the past two years.
• As agentic AI systems demand constant, low-latency access to huge datasets, storage is shifting from a passive repository into an active layer that determines whether AI deployments are profitable.