Cisco expands Splunk AI with NVIDIA on-prem stack and agent cost controls
The AMW Read
Incremental Cisco/Splunk product update that meaningfully advances on-prem agentic observability and SOC packaging with NVIDIA and AWS, at segment-level impact for enterprise agent adoption.
Cisco expands Splunk AI with NVIDIA on-prem stack and agent cost controls
At Splunk.conf in Denver, Cisco announced Cisco AI POD for Splunk inside Cisco Secure AI Factory with NVIDIA, bringing Splunk AI to self-managed on-premises, private-cloud, and air-gapped environments as of September 15, 2026. The pre-validated stack combines Cisco infrastructure, a new AI runtime, Kubernetes, and NVIDIA accelerated computing, with day-one delivery support from Accenture, bitsIO, Wipro, and World Wide Technology. Customers can self-host models such as Cisco Deep Time Series Model, Google Gemma 4, and OpenAI GPT-OSS 20B without sending data off-site, with NVIDIA Nemotron support planned in coming months. Splunk AI Assistant is available now; Agent Launchpad is due later in 2026 for ad-hoc investigation and custom agents, including agentic SOC use cases. Cisco also added Tokenomics to Splunk Agent Observability—now on Splunk Observability Cloud and Cisco Cloud Control—to track real-time token spend and attribution for AI agents and coding tools such as Claude Code, Codex, and Cursor, using Cisco Deep Time Series Model for spend forecasts. Separately, Splunk and AWS signed a multi-year joint-development pact to harden agentic SOC products against AI-speed attacks.
The bottleneck for enterprise agent rollout is no longer raw model capability; it is whether teams trust the system with sensitive machine data, can keep workloads on-prem for sovereignty, and can see cost before the invoice. Cisco is placing Splunk AI on infrastructure customers already trust, while NVIDIA supplies the accelerated layer that makes self-hosted agentic security and observability practical. Token-level visibility for coding agents turns a diffuse AI bill into an operational control plane, tying agent usage to budget and business outcomes—exactly the governance gap that has slowed large-scale agent deployment.
For builders and investors, the signal is distribution plus residency: pre-validated AI PODs and SI partners may become the default path for regulated agentic SOC rollouts, while the AWS multi-year build tests whether Splunk’s telemetry depth plus hyperscaler scale can outpace pure-play security-AI vendors on detect-investigate-respond speed.
