Virtana adds HPE AI Factory support for observability

Virtana has added support for HPE AI Factory through validation in the HPE Server Partner Program, extending its AI observability product to HPE AI-ready server infrastructure.

The addition is Virtana's fourth major AI Factory Observability integration this year, following similar tie-ups with Dell, Nutanix and AWS Bedrock. It reflects the company's effort to broaden coverage across large AI infrastructure environments as businesses invest in systems for training and inference workloads.

Validation through the HPE program allows customers to use Virtana's platform in production HPE AI environments, with confirmation that it interoperates with HPE servers. The software correlates GPU performance with memory, power, thermal and reliability data across HPE AI-ready servers.

The announcement comes as AI infrastructure operators face growing pressure to improve efficiency in systems built around expensive graphics processors. Virtana's research found that 41% of practitioners report GPU inefficiency and contention as AI workloads create scaling demands that older monitoring tools cannot explain.

Operational focus

This has become a central concern for companies building internal AI platforms, where utilisation rates, energy use and hardware reliability directly affect costs. Observability vendors are increasingly positioning their tools as a way to give infrastructure teams a clearer view of how workloads behave across servers, data pipelines and supporting systems.

For HPE-based deployments, Virtana said the platform tracks GPU and compute performance, including utilisation, memory consumption and throughput. It also monitors power draw and thermal conditions, flags signs of hardware degradation and follows data movement across the wider AI environment.

These measures are intended to help users distinguish between infrastructure faults, workload issues and orchestration problems. The platform also aims to link infrastructure consumption trends with capacity planning and workload costs.

Rather than relying on estimates, operators want ways to measure whether AI systems are running efficiently once they move beyond pilots into production. The validation is intended to address that concern in HPE environments, where organisations want assurance that third-party monitoring software will work as intended.

Paul Appleby, Chief executive officer of Virtana, set out the company's view of the shift in enterprise infrastructure management.

"AI is becoming foundational infrastructure for every industry, and the organizations that lead will be the ones that operate it with the same discipline they apply to every other mission-critical system," Appleby said.

He added: "The HPE AI Factory provides the foundation for enterprise AI. Virtana extends that foundation with end-to-end agentic observability, giving customers the visibility, governance, and operational control they need to optimize performance, improve efficiency and scale AI with confidence."

Broader build-out

The HPE integration is part of a wider push by Virtana to establish itself in AI infrastructure monitoring as demand grows for software that can span hybrid, on-premises and cloud systems. Vendors in this market are trying to show that conventional IT monitoring is not sufficient for AI workloads, which can create uneven resource demand and place sustained pressure on GPUs, networking and cooling.

Large enterprises have been spending heavily on AI hardware, but many are still working out how to govern those environments once they are live. Monitoring power and thermal conditions has become especially important as denser server deployments increase both operating costs and the risk of disruption.

Against that backdrop, vendor validation programs have taken on greater significance. They can help reassure buyers that tools will work in production settings and reduce the risk of operational problems in environments where failures may be costly.

Virtana said the HPE validation covers AI-ready server infrastructure used in enterprise AI deployments. Its platform is designed to provide visibility across applications, services, data pipelines, GPUs, CPUs, networks and storage, reflecting the growing need for tools that can track performance across multiple layers of an AI stack.

Virtana argues that as businesses deploy more agentic AI workloads across distributed infrastructure, observability needs to move beyond simple system monitoring into cross-stack analysis. For customers, the practical test will be whether such tools can reduce idle capacity, identify bottlenecks early and make AI systems easier to run at scale.

Its survey finding that 41% of practitioners report GPU inefficiency and contention underlines the scale of that challenge.