# WEKApod® **The World's Densest AI Storage & Memory System** AI changed the data center. We changed the math on every rack. ## WEKA-Built From the Chassis Up to Make Every U Count ### Dense by Design **Give Your GPUs More Room to Breathe** WEKApod packs up to 70 NVMe drives or up to 4 servers into a single 2U chassis, putting more storage and more compute in every rack unit. ### Mixed Flash Intelligence **QLC Economics With TLC Speed** Analyze workload characteristics in real time and automatically distribute data across TLC and QLC flash drives with WEKApod Prime’s AlloyFlash™ technology, which eliminates cache hierarchies and data movement, maximizing performance and cost efficiency. ### Backplane-Free Resilience **A Drive Fails. Nothing Flinches.** WEKApod isolates drives into independent failure domains. A cable-based interconnect replaces the shared backplane, so a drive failure stays contained and never reaches another domain. ### Thermal-Aware Power Throttling **Engineered to Stay Up When the Heat Is On** WEKApod is designed to keep running under thermal stress. Automatic power throttling within its 35°C ambient envelope reduces NVMe drive power as needed, keeping the system running rather than shutting down. ### Day-Zero Deployment **Ready to Run Before It Ships** WEKApod arrives ready to run. It ships factory-installed and pre-configured for drive count and capacity, with a guided, tool-free workflow. ## Industry-Leading Capacity & Performance - **1.1 EB** *Effective capacity per rack* - **10.2 TB/s** *throughput per rack* - **210 Million** *IOPs per rack* ## Customer Story > As sovereign AI adoption accelerates across ASEAN, governments and enterprises require infrastructure that is secure, compliant and built to perform at scale. The real bottleneck is how quickly data can reach GPUs, making high-speed data architecture just as critical as the GPUs themselves. By partnering with WEKA, we are strengthening our sovereign AI platform, bringing together trusted cloud infrastructure, advanced data capabilities and network intelligence, so that customers can deploy complex AI workloads with confidence. — Bill Chang, CEO, Singtel : REAI ## Frequently Asked Questions ### What is WEKApod? WEKApod is WEKA's AI storage and memory appliance: a family of systems built on hardware WEKA designed from the chassis up, running WEKA NeuralMesh, WEKA's software platform for AI storage and memory. It ships in three configurations tuned for capacity, price-performance, and throughput. Based on published system specifications, the latest WEKApod systems deliver the highest performance and capacity density on the market today. ### What's the difference between WEKApod and WEKA NeuralMesh? WEKA NeuralMesh is WEKA's AI storage and memory platform software. WEKApod is the hardware WEKA built to run it. NeuralMesh runs on WEKApod, on customer-owned hardware, or converged on GPU servers through NeuralMesh Axon. WEKApod delivers the software's full performance on hardware designed for it, with no configuration required. ### What's unique about WEKApod? WEKApod runs on a chassis WEKA designed from scratch. A single two-unit (2U) chassis holds up to 70 NVMe drives and up to four independent servers. A cable-based, backplane-free interconnect creates up to four independent failure domains per chassis, so a single drive event stays contained. A PCIe Gen 6 internal fabric gives every drive full bandwidth at once. WEKA NeuralMesh manages thermal stress in software, keeping WEKApod running at 35°C ambient, the real conditions of a production AI datacenter. ### Which WEKApod configuration is right for me? Prime Max maximizes capacity density, up to 1.1 effective exabytes per rack. Prime balances capacity and throughput with the best price-performance in the family. Nitro maximizes performance, up to 210 million IOPS per rack. All three run the same WEKA NeuralMesh software. ### Can WEKApod be added to an existing WEKA NeuralMesh deployment? Yes. WEKApod joins an existing WEKA NeuralMesh cluster as a standard node, using the same software and the same control plane. Customers already running NeuralMesh can add WEKApod nodes to the same cluster and get the density advantage without changing how they operate. ### How does WEKApod fit so much capacity and performance into one rack? WEKA engineered WEKApod from the chassis up instead of adapting an OEM server platform. A cable-based, backplane-free drive interconnect replaces the physical backplane that normally limits how many drives fit in a chassis. Removing that constraint frees the space a backplane would take, enabling WEKApod to pack up to 70 NVMe drives into two units. ### How does WEKApod pack multiple servers into the same chassis? The backplane-free interconnect that frees up drive slots also decouples drives from any single server. This allows WEKA to wire up to four independent servers into the same two-unit chassis, each with its own connection into the shared drive pool, instead of walling off a dedicated set of drives per server. Compute density scales in the same footprint as storage density because servers aren't competing for siloed storage. ### What happens when a drive fails in WEKApod? WEKApod connects every drive independently through a cable-based interconnect instead of a shared backplane. That design creates up to four independent failure domains in a single two-unit (2U) chassis. A single drive event stays contained in its own domain, with no cascade to other drives and no need for cluster-level intervention. ### How does WEKApod's internal architecture deliver full bandwidth to every drive? WEKApod runs on a PCIe Gen 6 internal fabric, the highest-bandwidth interconnect available today. Every drive gets full bandwidth at the same time, with none waiting behind another for access. Performance scales with drive count instead of hitting an internal bottleneck. ### How does WEKApod keep running under thermal stress? WEKA NeuralMesh monitors chassis conditions continuously and manages thermal stress in software. When temperatures rise, NeuralMesh reduces NVMe drive power instead of triggering a shutdown, so the system keeps running and performance degrades gracefully. WEKA built WEKApod to operate at 35°C ambient, the real conditions inside a production AI datacenter. ## Related Resources - [When Everything Is Scarce, Density Is the Only Lever You Have](/article/when-everything-is-scarce-density-is-the-only-lever-you-have) (Article) - [The Capability Stack Behind Exabyte-Scale AI](/article/the-capability-stack-behind-exabyte-scale-ai) (Article) - [WEKApod. More AI. Same Data Center. No Compromise](/resources/solution-brief/wekapod-more-ai-same-data-center-no-compromise/) (Solution Brief) - [WEKApod: Next-Generation AI Storage Appliances](/resources/datasheet/wekapod-next-generation-ai-storage-appliances/) (Datasheet)