Cisco Expands Splunk AI to On-Premises Environments via NVIDIA Partnership

The initiative marks an expansion of the partnership between Cisco and NVIDIA, targeting customers who require strict data sovereignty and cannot migrate sensitive information to public clouds. By utilizing Cisco infrastructure alongside NVIDIA accelerated computing and Kubernetes-based architecture, teams can now host generative AI models—including Google Gemma 4 and Cisco’s Deep Time Series Model—directly within their own data centers. This deployment model supports Splunk’s existing AI Assistant and upcoming Agent Launchpad, facilitating local agentic investigations without external data exposure.

Beyond infrastructure, Cisco is tackling the financial uncertainty associated with AI scaling through new observability features. The introduction of Tokenomics within Splunk Agent Observability provides real-time tracking of AI token consumption and agent performance. This tool allows security and IT teams to monitor costs and enforce runtime guardrails, preventing unsafe actions like data leaks or hallucinations. Furthermore, a multi-year agreement with AWS will see the companies co-developing security solutions designed to counter AI-driven threats, pushing the development of an autonomous, agentic security operations center capable of defending infrastructure at machine speed.