NeuralMesh™ by WEKA®

NeuralMesh by WEKA is the world's only storage system purpose-built for accelerating AI at scale. WEKA's software-defined microservices architecture does not just handle scale, it gets faster, more efficient, and resilient at exabytes and beyond. It's designed for ultimate flexibility, adapting effortlessly to your deployment strategy from bare metal to Azure Blob. NeuralMesh by WEKA was engineered to eliminate bottlenecks, delivering consistent microsecond latency, scalable resilience, and container-native flexibility across any deployment model.

At its core, WEKA is built on an innovative architecture that accelerates AI model training by 10x while achieving 93% GPU utilization, dramatically reducing time-to-insight for complex workloads.

WEKA combines dense NVMe storage of Azure Virtual Machine Lsv3-series instances with Azure Blob Storage in a single, efficient namespace, for your high-performance workloads, scaling to billions of files and hundreds of petabytes. It has a rich feature set that includes transparent object tiering, instantaneous snapshots, snap-to-object (remote clouds), backup, disaster recovery (“DR”), encryption, quotas, Active Directory integration, Kubernetes CSI driver, and much more.

NeuralMesh by WEKA Cloud Edition:

-License for the quantity of an all-flash cluster, built using the NVMe storage on Lsv3-series Azure Virtual Machines*

-POSIX, NFS, SMB, and S3 protocol support

-Shared namespace that transparently combines instance-based performance storage with high-capacity Azure -Blob Storage to deliver the best overall application performance with the best economics

-Snapshot to Azure Blob Storage* provides a low-cost way to protect critical data with fast recovery times for fault tolerance in the event of an availability zone failure

-Incremental snapshots that can be scheduled to a minute granularity

-Snapshots to Azure Blob Storage can enable migration to another cloud region

-XCL-Cloud license can be used for the total quantity of flash storage used across any number of clusters, regardless of cloud region

NeuralMesh by WEKA is particularly valuable for:

-Drug Discovery: Accelerate cryo-em workflows from weeks to hours

-Genomics: Reduce genomic processing workflows from months to days

-Content Production: Achieve 120 frames per second rendering in cloud environments

-AI Foundation Model Training: Train models at exabyte scale while maintaining terabyte-level costs

-AI inference at Scale: Sub-millisecond latency at petabyte scale to drive responsive AI applications

-HPC Organizations: Execute complex scientific simulations with unprecedented efficiency

-HiTech and EDA: 30x improvement in checkout times and support full chip and semiconductor design

*NOTE: Lsv3-series Azure Virtual Machine instances and Azure Blob Storage cost is not included in the WEKA Data Platform licensing.

Pricing is a starting point and is discounted based on total consumption and committed term. Contact orders@weka.io to discuss private contracts and custom prices.