Cohesity at Cisco Live 2026: Protecting and activating data for sovereign AI

In February, we announced our on-premises AI solution with Cisco, built on a clear premise: enterprises can secure and activate their data for AI without weakening governance. At NVIDIA GTC 2026, we jointly showcased how Cohesity Gaia—powered by NVIDIA AI Enterprise—turns governed enterprise data into the foundation for deeper agentic reasoning.

Since 2018, Cohesity and Cisco have worked together to help enterprises protect, manage, and recover data on Cisco UCS. The next chapter builds on that proven data resilience foundation: customers can continue protecting and governing their critical data, then extend that same foundation with Cohesity Gaia to unlock more value from their data for AI. 

Now, at Cisco Live 2026, we’re bringing this vision full circle: helping organizations to activate on-prem enterprise data for AI—without moving it out of their environment.

Enterprise data is under attack–and AI is raising the stakes

Ransomware, insider threats, and destructive cyberattacks hit production and backup data every day. For regulated industries—healthcare, financial services, public sector—a single incident can take down operations, impact compliance, or erase years of institutional knowledge.

AI can increase these risks. Sensitive data powering AI systems flows across pipelines, vector stores, and agent workflows, expanding the attack surface. At the same time, many regulated organizations cannot move sensitive data into cloud-native AI platforms at all—limiting their ability to adopt AI without compromising compliance. To remain secure and confident in the AI era:

• Production data must be protected and recoverable at scale across every workload.
• The data feeding AI must remain governed and policy-controlled.
• AI itself must run where the data lives, particularly in regulated and sovereign environments where moving sensitive data to the cloud can violate residency and compliance requirements.

Together, Cohesity, Cisco, and NVIDIA deliver a comprehensive approach that addresses all three imperatives.

The foundation: Cyber resilience on Cisco’s AI infrastructure

Before enterprise data can power AI, it has to be trusted—protected, governed, and recoverable. 

Cohesity Data Cloud running on Cisco's AI-ready infrastructure provides that foundation as part of Cisco’s Secure AI Factory. It’s a hardened, production-proven platform based on zero-trust principles that protects enterprise data across environments—from VMs and databases to NAS, SaaS, and identity systems.

Cisco brings AI-ready compute, fabric, security, and operations into a single, policy-driven platform. With Cisco’s unified infrastructure management platforms and hardware-rooted trust, customers can extend the same operational and security model they use for resilient production workloads into their AI environments. 

Built for customer-controlled environments—including sovereign and air-gapped deployments—Cohesity Data Cloud on Cisco UCS supports protection, recovery, and governance across enterprise data environments. Here’s how Cohesity delivers across each stage:

• Protect: Global governance across 1,000+ data sources, with unified policies, SLAs, and reporting.
• Strengthen: Immutable cyber vaults and a platform built on zero-trust principles to help ensure data remains recoverable even when production environments are compromised.
• Respond: Continuous threat detection designed to integrate with SIEM and SOAR workflows.
• Practice: Clean room recovery with Instant Mass Restore can help teams rehearse incident response and recover petabytes of data in hours, not weeks.
• Optimize: 230+ data classifiers and DSPM integrations help strengthen data risk posture and surface unprotected data before attackers do.

Customers already rely on this platform to defend Active Directory, recover VMs at production scale, and restore operations after ransomware attacks. It runs on Cisco compute, with Cisco zero-trust networking and hardware-rooted security at its foundation.

As AI agents become critical enterprise applications, their configurations, memory, and workflows must also be protected and recoverable.

Cohesity Data Cloud extends cyber resilience to AI workflows by rolling back compromised embeddings, recovering from agent errors, isolating AI incidents in clean rooms, and helping preserve immutable source-of-truth data even if downstream systems are affected. With that resilient foundation in place, organizations can securely activate governed enterprise data for AI workflows.

That’s where Cohesity Gaia comes in.

What's new: Gaia on Cisco AI-ready infrastructure, powered by NVIDIA

Most enterprise data remains on-prem and largely inaccessible to AI systems. Cloud-native AI tools force organizations into difficult tradeoffs: move sensitive data to the cloud, limit AI to partial datasets, or build complex custom integrations.

Gaia Self-Managed with Cisco UCS provides a production-ready, turnkey path to run sovereign AI on governed on-prem data. It runs entirely inside the customer environment on Cisco UCS—powered by NVIDIA AI Enterprise and accelerated by NVIDIA Blackwell GPUs:

• Gaia delivers AI-ready governed data to agents: Semantic search and summarization, RBAC and file-level permissions, multilingual support, and RAG workflows running on Cisco UCS with NVIDIA Blackwell and Hopper GPUs (NVIDIA RTX PRO 6000 BSE, NVIDIA H100, NVIDIA H200) powering inference, vectorization, and semantic retrieval.
• NVIDIA AI Enterprise powers core AI capabilities within Gaia’s AI engine: NVIDIA NIM microservices run the LLM securely inside the customer's environment, an NVIDIA Nemotron reranking NIM is designed to improve answer accuracy, and NVIDIA NeMo Guardrails help keep responses policy-aligned.
• Cisco infrastructure extends enterprise security controls into sovereign AI environments: enabling workload isolation and hardware-rooted trust.

In many environments, deployment can be completed in weeks, depending on infrastructure and configuration, while keeping data on-prem under existing governance controls. Gaia provides a path to build on the protected data foundation they already manage, rather than creating a separate AI stack or moving sensitive data into cloud-native tools.

Deployment options align with regulated enterprise deployment patterns:

• Self-managed deployments allow organizations to run Gaia within customer-controlled environments while integrating with agentic AI tools like Microsoft Copilot, Google Gemini—with Glean and more integrations coming soon.
• Air-gapped deployments support environments where AI and data must remain fully isolated.

The business case for sovereign AI with Cohesity, Cisco, and NVIDIA

A common misconception with sovereign AI is that enterprises trade performance for governance. With Gaia Self-Managed from Cohesity, Cisco, and NVIDIA, sovereign AI can, for many workloads, outperform cloud-native AI implementations.