Cisco AI POD for Splunk Brings GPU-Accelerated AI On Premises

Cisco and NVIDIA used Splunk .conf26 in Denver to announce general availability of Cisco AI POD for Splunk, a pre-validated, self-managed platform for bringing Splunk AI into customer-controlled environments. The offering is aimed at organizations that need to apply AI to sensitive machine data without moving that data to a public cloud. 3

What Cisco AI POD for Splunk is

Cisco AI POD for Splunk is a deployment configuration within the Cisco Secure AI Factory with NVIDIA reference architecture. It combines Splunk AI runtime software, Cisco infrastructure, and NVIDIA-accelerated computing for Splunk AI workloads. Customers can deploy it on premises, in private-cloud environments, or in air-gapped environments. 3

The central proposition is data control. Rather than exporting logs, telemetry, alerts, and operational records to an outside AI service, enterprises can run Splunk AI where their machine data is already retained. Cisco and NVIDIA describe the platform as self-managed for Splunk Enterprise customers. 3

That makes the offering particularly relevant to public-sector organizations, regulated industries, and globally distributed enterprises with data-residency, sovereignty, or security constraints. 13

What data and workflows it targets

The platform is intended for AI workflows across the operational data commonly handled in Splunk, including:

• Security telemetry and investigation data
• Infrastructure, network, and device signals
• Application and service-health data
• Logs, alerts, and other operational event data 38

Cisco’s stated use cases include AI-assisted security, observability, and operations. In particular, Splunk AI Assistant is positioned for ad hoc agentic investigations, while the wider .conf26 program emphasized AI-powered workflows spanning security operations and observability. 35

Software, infrastructure, and deployment model

The announcement describes a Kubernetes-based architecture for deploying the Splunk AI runtime on customer-controlled infrastructure. Independent reporting on the announcement identifies Red Hat OpenShift as part of the turnkey stack alongside Cisco UCS compute, NVIDIA GPU acceleration, and Splunk’s AI runtime. 310

This arrangement is meant to give customers an alternative to sending operational data to externally hosted AI platforms. Organizations retain responsibility for operating the environment, but receive a pre-validated hardware-and-software configuration rather than assembling every layer themselves. 3

Cisco says the runtime can support self-hosted AI capabilities on customer infrastructure. It also named Accenture, bitsIO, Wipro, and World Wide Technology as implementation partners. 3

What is available now—and what remains unclear

Splunk AI Assistant is available with the new platform. Cisco also points to Agent Launchpad, intended to help customers build custom AI agents, as part of the broader on-premises Splunk AI direction. 317

However, the available announcement material does not provide a complete catalog of supported foundation models, nor does it establish support for specific NVIDIA, open-source, or proprietary models. Customers evaluating the platform should therefore confirm model availability, hosting options, and operational requirements directly with Cisco or their deployment partner. 317

How the AI POD fits Cisco’s broader strategy

Cisco’s larger goal is to turn Splunk into a data-and-action layer across network operations, security, observability, and AI infrastructure. In this framing, the AI POD supplies the on-premises infrastructure needed to run trusted AI workflows at production scale, rather than treating AI as a standalone cloud experiment. 38

The related Cisco Data Fabric, powered by the Splunk Platform, is intended to let organizations analyze data where it resides, connect data across domains in real time, and transform machine data for model training and governed agentic action. 17

Together, the components point to a production-oriented architecture:

• Cisco AI POD for Splunk provides the validated, customer-controlled AI infrastructure.
• Splunk AI runtime and AI Assistant bring AI functions closer to Splunk-held machine data.
• Agent Launchpad extends the approach toward custom agents.
• Cisco Data Fabric is meant to connect and prepare distributed data for analysis and governed action. 317

The supplied evidence does not substantiate the specific relationship between the AI POD announcement and a claimed expansion of Native Splunk in Cisco Nexus One from single-node to multi-node architecture, or a particular NVIDIA-Certified Systems foundation. Those details should be treated as unverified in the context of this announcement.

Bottom line

Cisco AI POD for Splunk is a generally available path for Splunk Enterprise customers to run GPU-accelerated, self-managed AI in their own on-premises, private-cloud, or air-gapped environments. Its practical value is strongest for organizations that need AI over security and operational machine data while preserving control over where that data resides. 3

The announcement is less a claim about a specific model than an infrastructure strategy: combine Splunk AI software, Cisco-operated infrastructure, and NVIDIA acceleration so enterprises can move governed AI and agentic workflows from pilots toward production without requiring sensitive data to leave their environment. 317