WEKA has introduced NeuralMesh™, a software-defined storage architecture that it says provides microsecond data delivery for real-time agentic AI. Chief executive Liran Zvibel called the mesh design “the intelligent foundation for exascale reasoning.”
NeuralMesh links data, compute and AI services in a self-healing, containerised fabric that becomes faster and more resilient as capacity grows from petabytes to exabytes, rebuilding failed hardware in minutes. Chief product officer Ajay Singh said the system “unleashes GPUs and TPUs, driving 93 percent utilisation at Stability AI and cutting storage cost by 20 percent.”
Limited release begins today for enterprise and hyperscale deployments; general availability is scheduled for autumn 2025.
