Cisco Unveils AI-Powered Data Fabric to Revolutionize Machine Data Management

Cisco aims to turn machine data chaos into AI Intelligence with Data Fabric - SiliconANGLE

Cisco aims to turn machine data chaos into AI Intelligence with Data Fabric Continuing its campaign to position itself as the control plan for the artificial intelligence era, Cisco Systems Inc. today is using its Splunk subsidiary's .conf25 to unveil the Splunk-powered Cisco Data Fabric, a network architecture designed to unify stovepipes of fragmented machine data. Cisco also unveiled Splunk Federated Search for Snowflake. This integration enables users to access Snowflake data directly from within the Splunk interface, combining operational machine datasets with business data. The fabric is based on the Time Series Foundation Model, an AI engine optimized for time-series analytics and forecasting on vast amounts of data. Cisco said the fabric can be used for anomaly detection, forecasting and automated root-cause analysis to enable proactive operations and rapid incident response. Ingested data is contextualized and turned into actionable insights in real time to accelerate decision-making. Another pillar of the fabric is a data normalization layer called the Unified Intelligent Data Foundation. It consolidates data flows from security, operations, development and networking to enable more efficient cost management. Layered atop it is Borderless Real-Time Search, which federates queries across data sources such as Amazon Web Services Inc.'s S3, Apache Iceberg, Databricks Inc.'s Delta Lake, Snowflake Inc.'s Data Cloud and Microsoft Corp.'s Azure to route workloads optimally to the appropriate compute engine. The Splunk Machine Data Lake provides a persistent substrate for model training and enterprise analytics. Together with the Splunk AI Toolkit and MCP server, it handles AI-native services for onboarding, monitoring and self-healing operations. Cisco said the architecture is open and adaptable. Federated Search can be use for unlimited onboarding of Snowflake as a data source, with Splunk Search Processing Language queries executed across Snowflake datasets. Users can join Splunk and Snowflake data with query execution automatically distributing workloads between Snowflake's analytics and Splunk's search layers. Cisco has struggled to articulate a compelling story for the burgeoning AI market, but the control plane strategy outlined at Cisco Live 2025 in June was "the first time in 15 years that Cisco has shown a truly integrated roadmap instead of a stitched-together portfolio with several acquisitions on slideware," theCUBE Chief Analyst Dave Vellante wrote following the event. No longer just selling boxes, Cisco is now positioning itself as an integrated platform spanning network, compute, security and analytics, bolstered by Splunk's data fabric, Vellante observed. Cisco's ambitions to unify data, apply AI analytics in real-time, search across distributed stores, and feed AI training pipelines address the pain enterprises are experiencing as they wrestle with a flood of machine data, ranging from logs to telemetry to event streams. The federated Snowflake search also demonstrates Cisco's willingness to integrate with rather than replace incumbent enterprise data platforms as it has tried to do in the past.

Cisco's Patel: 'Splunk Is The Machine Data Fabric For The AI Era'

In a keynote at the Splunk .conf25 event Monday Cisco President and Chief Product Officer Jeetu Patel unveiled the new Cisco Data Fabric architecture, based on the Splunk platform, and touted Splunk's technology as a way to eliminate the infrastructure, trust and data gaps that hinder AI adoption. Splunk's software for collecting and managing huge volumes of machine data will play a major role in filling the infrastructure, data and trust gaps that are holding back broader use of AI and agentic AI, a Cisco Systems top executive said Monday. In an opening keynote at the Splunk .conf25 customer and partner event in Boston Monday evening Cisco President and Chief Product Officer Jeetu Patel outlined a vision for how the combination of Cisco and Splunk technology -- including the newly introduced Cisco Data Fabric -- will help businesses and organizations move into the agentic AI era. "We are now squarely moving into the second era of AI, which essentially is agents being able to conduct jobs and tasks for us almost fully autonomously," Patel said. "And this is going to be much more about automation of workflows, rather than just individual productivity," he said, contrasting agentic AI with the focus on generative AI and chatbots that have marked the first years of the AI revolution. [Related: Cisco CEO: Record AI Infrastructure Orders Highlight 'Undeniable Capability And Relevance' Of Cisco Portfolio] Patel noted that major shifts in IT platforms seem to happen every 10 to 15 years, but the AI wave in general, and agentic AI in particular, "is one that I don't think we have seen in our lifetimes. And all of this is going to have massive implications for technology architectures." Network and security giant Cisco Systems acquired Splunk for $28 billion in March 2024. Since then, Cisco, while continuing to operate Splunk as a subsidiary, has been integrating its product portfolio with Splunk's platform for collecting and managing machine data and its software, such as observability and security management tools, that leverage that data. Patel, acknowledging pleas of "Please don't screw up Splunk" from customers and partners in the wake of the acquisition, said Cisco has continued to leverage Splunk as an innovation engine. And the announcements at this week's Splunk .conf25 are the latest results of that approach. The Cisco executive said AI faces three constraints today. One is infrastructure demands driven by agentic AI, which he said puts more sustained, continuous demands on IT capacity than the processing "spikes" created by the inferencing engines that drive generative AI and chatbots. "You're going to see a very different pattern of infrastructure demand in the future compared to what you've had in the past," he said. What Patel called "the trust deficit" is another constraint. AI systems can be black boxes: "If you can't see something up and down the [IT and data] stack, you're not going to be able to trust it. You have to have visibility to be able to trust the system...from silicon to agent," he said. Another aspect of the trust deficit is the sometimes-suspect safety and security of AI systems, Patel said. That includes the unpredictability of the AI models themselves -- the potential for "hallucinations or toxicity" -- and the possible new security "threat vectors" such as prompt injection attacks that AI systems create. And the "third gap" is agentic AI systems' need for greater volumes and varieties of data, Patel said. "Data is the essential fuel that unlocks AI," he said. "The good news is that Cisco is actively working on every single one of these challenges," Patel said, and Cisco -- in conjunction with Splunk -- "is going to be the critical infrastructure for the AI era." Patel said Cisco and Splunk have been integrating their product portfolios to address the infrastructure and trust gaps, providing technology for "AI-ready data centers" such as secure global connectivity and digital resilience. At one point the giant screen behind Patel scrolled down through a list of dozens of Cisco-Splunk integrations since the acquisition, one example being the linkage of Cisco AI Defense with the Splunk platform. Onstage Splunk executives demonstrated the upcoming integration of Cisco AI Canvas, the agentic AI system for IT operations that the company unveiled at Cisco Live in June, with the Splunk Cloud Platform. In another example Patel drew cheers from the audience when he announced that Splunk customers can now ingest security data from Cisco firewall systems at no charge. "The future of the SOC [security operations center] is agentic," he declared. Much of Patel's keynote focused on addressing the AI data gap. AI systems have generally been trained on and use human-generated information such as news stories, historical texts and literature. The executive said the potential to use the kinds of machine data collected and managed by Splunk to expand AI systems' training and capabilities is "a massive, untapped opportunity." That's the thinking behind the new Cisco Data Fabric, an architecture incorporating Splunk Enterprise and Splunk Cloud Platform, that Cisco said will reduce the cost and complexity of handling machine data at scale and leveraging that data for AI applications such as training custom AI models and powering agentic workflows. "Splunk is the machine data fabric for the AI era," Patel said. Patel said Cisco Data Fabric will process machine data "at ludicrous scale," unlock proprietary data for AI, and provide a unified experience for people and AI agents. The system will also be capable of correlating multiple machine and business data streams to better drive AI systems. And it can transform data across on-premises, cloud and edge systems, including ITOps, SecOps, DevOps and NetOps, into real-time, actionable insights, according to the Cisco Data Fabric announcement. As part of the Cisco Data Fabric, Cisco and Splunk are developing a time series data foundation model, slated for availability in November through the Hugging Face community, that can provide time-series data for AI agents and assistants. The companies are also developing a machine data lake system that's planned for general availability in February 2026. "We're going to be teaching AI to speak machine data," Patel said. "AI has [already] learned how to speak natural language. We're going to teach it to speak machine data. AI agents are going to be multi-lingual. Splunk is the machine data fabric for the AI era." Also unveiled at the opening of Splunk .conf25 was Splunk Federated Search for Snowflake, a new Splunk platform integration that allows users to connect, query and combine operational and business data across both Splunk and the Snowflake Data Cloud. That capability is scheduled to be generally available in February 2026.