WEKA releases next-generation WEKApod storage appliances optimized for AI workloads to improve storage infrastructure economics.
From SC25, WEKA announced the next generation of its WEKApod appliances designed to address the performance-versus-cost trade-offs facing AI infrastructure deployments. The completely redesigned WEKApod Prime appliance achieves 65% better price-performance by intelligently placing data across mixed flash configurations, while WEKApod Nitro doubles performance density with refreshed hardware, enabling organizations to accelerate AI and high-performance computing innovation, maximize GPU utilization, and serve more customers. WEKApod appliances deliver NeuralMesh by WEKA—the world's only storage system purpose-built to accelerate AI at scale—in ready-to-deploy, pre-validated configurations with improved setup utilities offering a plug-and-play experience. Organizations can start with as few as eight servers and scale to hundreds as their deployments grow, accessing NeuralMesh's complete enterprise feature set including distributed data protection, instant snapshots, multi-protocol access, automated tiering, encryption, and hybrid cloud capabilities.
WEKApod Prime offers a unique approach to mixed-flash technology through its AlloyFlash innovation, intelligently combining TLC and eTLC flash drives in configurations supporting up to 20 drives in a 1U rack or 40 drives in a 2U rack configuration. Unlike traditional tiered storage solutions that introduce cache hierarchies and data movement penalties, WEKApod's AlloyFlash delivers consistent performance across all operations, including full write performance without throttling, while achieving breakthrough economics. The appliance already is being used by leading AI cloud pioneers like the Danish Centre for AI Innovation (DCAI).
This approach yields exceptional density gains directly addressing datacenter resource constraints: 4.6x better capacity density, 5x better write IOPS per rack unit versus the previous generation, 4x better power density at 23,000 IOPS per kW or 1.6 PB per kW, and 68% less power consumption per terabyte while maintaining the extreme performance AI workloads demand. WEKApod Nitro is purpose-built for AI factories running hundreds or thousands of GPUs, delivering 2x faster performance and 60% better price-performance through upgraded hardware including NVIDIA ConnectX-8 SuperNIC delivering 800 Gb/s throughput and 20 TLC drives in a compact 1U form factor. Both configurations are turnkey NVIDIA DGX SuperPOD and NVIDIA Cloud Partner-certified appliances, eliminating weeks of integration work and enabling organizations to bring customer services to market in days rather than months.
Nadia Carlsten, CEO of the Danish Centre for AI Innovation, stated: "Space and power are the new limits of innovation in data centres. WEKApod's exceptional storage performance density allows us to deliver hyperscaler-level data throughput and efficiency within an optimised footprint—unlocking more AI capability per kilowatt and square metre. This efficiency directly improves economics and accelerates how we bring AI innovation to our customers." Ajay Singh, Chief Product Officer at WEKA, noted: "AI investments must demonstrate ROI. WEKApod Prime delivers 65% better price-performance without compromising on speed, while WEKApod Nitro doubles performance to maximize GPU utilization. The result: faster model development, higher inference throughput, and better returns on compute investments that directly impact profitability and time-to-market."