Inside the Efficient AI Factory | FIRMUS x WEKA

How FIRMUS builds and operates energy-efficient AI factories with WEKA, from flexible storage to record-setting benchmark runs — https://www.weka.io/customers/firmus FIRMUS designs, builds, deploys, and operates AI factories with a singular focus: the most efficient AI infrastructure possible, optimized from model to grid. In this customer story, the FIRMUS team explains how WEKA fits into that mission - giving them the flexibility to serve customers who need block storage, S3, or both, and the software-defined speed to bring new clusters online fast on certified hardware. They also share how running the NeuralMesh™ AMG benchmark pushed the limits of GPU, network, and storage scaling, unlocking large KV cache capacity for inference workloads with big context windows. The result: efficiency at every layer, lower token costs, and the ability to engineer out obsolescence by making existing GPU systems ready for the next generation of workloads. *CHAPTERS* 0:00 The FIRMUS mission: efficiency from model to grid 0:36 Flexible storage: block and S3 on one platform 0:55 Fast, software-defined deployment 1:15 Running the NeuralMesh AMG benchmark 1:41 KV cache capacity and large context windows 1:59 Efficiency at every layer lowers token cost 2:09 Engineering out obsolescence 2:24 A partnership built on engineering 👉 *Connect with WEKA:* Website: https://www.weka.io?utm_source=youtube&utm_medium=social&utm_campaign=brand LinkedIn: https://www.linkedin.com/company/weka-io?utm_source=youtube&utm_medium=social&utm_campaign= X: https://x.com/weka?utm_source=youtube&utm_medium=social&utm_campaign=inference