Virtana announced AI Factory Observability for Dell AI Factory environments, bringing its AI Factory Observability platform to one of the industry’s most widely deployed enterprise AI infrastructure stacks. 

The integration spans Dell PowerEdge compute, PowerScale and ObjectScale storage, high-performance networking fabrics, including InfiniBand, Ethernet, and NVLink, and Dell’s Smart Fabric Manager (SFM) orchestration layer. As enterprises deploy Dell AI Factory to run GPU-intensive training and inference at scale, the operational challenge shifts from infrastructure acquisition to infrastructure performance: understanding not just whether components are running, but whether the system is producing outcomes efficiently. Virtana directly addresses this challenge, giving infrastructure and AI platform teams end-to-end visibility and control across every layer of the Dell AI Factory stack. Having established deep integrations with NVIDIA and Nutanix, Virtana continues to extend full-stack observability across the major ecosystem environments where enterprises are building and operating AI at scale.

“Dell AI Factory gives enterprises a world-class foundation for running AI at scale. The challenge every organization faces, regardless of platform, is connecting infrastructure performance to actual AI outcomes,” said Paul Appleby, CEO of Virtana. “Virtana solves that. We give Dell AI Factory customers the end-to-end visibility to know whether their GPUs are producing value, where constraints exist, and how to optimize the system to get more from their investment.”

Virtana AI Factory Observability integrates natively across every layer of the Dell AI Factory architecture. Rather than adding telemetry volume, Virtana connects signals across the entire stack and explains why the system behaves the way it does by correlating GPU performance with storage I/O, network fabric throughput, workload orchestration, and AI model output in a single operational view.

Virtana AI Factory Observability capabilities delivered across the Dell AI Factory stack include:

• GPU and compute performance across PowerEdge infrastructure map utilization to workload output, expose idle and misallocated capacity, and correlate GPU performance with upstream and downstream dependencies
• Storage observability across PowerScale and ObjectScale identify I/O latency that directly impacts training and inference, correlate data pipeline performance with model slowdown, and enable storage bottlenecks visible and actionable
• Network fabric intelligence across InfiniBand, Ethernet, and NVLink detect east-west congestion across GPU clusters, correlate fabric performance with job latency, and identify constraints that limit scaling efficiency in distributed training environments
• Cluster and fabric management visibility through SFM integration surface workload placement behavior and provide directional insight into potential imbalances or inefficiencies, without requiring deep manual correlation across tools
• Node-level hardware intelligence from iDRAC telemetry correlate power, thermal, and health signals with system impact to distinguish hardware issues from workload or orchestration problems
• AI workload and cost optimization connect LLM behavior, token usage, and latency to infrastructure performance, map cost per token to actual infrastructure consumption, and enable true optimization of AI economics

“AI workloads at scale are complex by nature; they span GPUs, storage, networking, and orchestration. Performance depends on how all of those layers interact,” said Amitkumar Rathi, Chief Product Officer at Virtana. “The Dell AI Factory gives enterprises a powerful, integrated foundation. Virtana connects the signals across that foundation so teams can resolve issues faster, maximize GPU ROI, and scale from pilot to production with confidence.”