Cisco and its Splunk unit have unveiled a Nvidia-accelerated AI POD purpose-built for Splunk workloads, together with a new Splunk Agent Launchpad and 'tokenomics' observability, extending agentic AI into on-premises, private-cloud and air-gapped enterprise environments.
An AI POD Sized For Splunk Data
The Cisco AI POD for Splunk bundles Cisco infrastructure, NVIDIA compute, an AI runtime and a Kubernetes architecture into a pre-validated appliance sized for Splunk workloads. It is available now and lets customers run Splunk AI Assistant, an agentic co-pilot, entirely behind their firewall, with Splunk Agent Launchpad following later in 2026 to let SOC teams build custom agents and run ad-hoc investigations without shipping data to a public cloud.
"One of the biggest roadblocks to enterprise AI today is that it's too hard to deploy," said Cisco president and chief product officer Jeetu Patel.
Tokenomics And Agent Observability Get A First-Class Home
The announcement also introduces Splunk Agent Observability with 'tokenomics,' tracking AI token spend and coding-agent usage in real time and forecasting consumption using Cisco's Deep Time Series Model. That capability lands directly in Splunk Observability Cloud and Cisco Cloud Control, giving finance and platform teams a way to attribute AI cost to specific agents, teams and business processes as adoption scales.
"Enterprises need to bring AI where their data lives, especially when security and sovereignty requirements require critical workloads to stay on-premises," said Justin Boitano, NVIDIA's vice president of enterprise AI.
Private-First Agentic AI Grows Louder
The Cisco/Splunk POD is one of several private-first pushes this week: OpenText and Cohere paired for sovereign agentic AI, Factory raised $200 million for enterprise coding Droids and DataCebo shipped SDV 2.0 for synthetic enterprise data. The through-line: regulated buyers will not send crown-jewel data to shared clouds, so vendors are packaging Nvidia GPUs, agent frameworks and observability into ready-to-deploy stacks.
Reporting based on Cisco Newsroom, Unite.AI, Solutions Review and SiliconANGLE.