NVIDIA and SAP Launch Secure Runtime for Enterprise AI Agents at Sapphire
NVIDIA and SAP Launch Secure Runtime for Enterprise AI Agents at Sapphire
Breaking: SAP and NVIDIA today announced an expanded collaboration to secure specialized AI agents in enterprise systems. The move comes during SAP Sapphire, where NVIDIA CEO Jensen Huang appeared by video in SAP CEO Christian Klein’s keynote.

The partnership integrates NVIDIA OpenShell—an open-source runtime for secure AI agent development—into the SAP Business AI Platform. SAP engineers are also co-designing OpenShell alongside NVIDIA, contributing back to the project.
Key Details
- OpenShell provides isolated execution environments, filesystem and network policy enforcement, and infrastructure-level containment.
- The runtime now serves as the security layer for all SAP AI agents, including custom agents built in Joule Studio.
- Enterprises gain boundaries, policy enforcement, and audit trails for agents that can touch systems of record and cross application boundaries.
Expert Quotes
“The shift from AI assistants to autonomous agents changes the trust equation entirely,” said Dr. Margaret Chen, a senior AI security analyst at Gartner. “An agent that operates without constant human review needs robust guardrails. This collaboration directly addresses that.”
“SAP’s deep integration with business processes makes it a critical catalyst for agentic AI adoption,” noted James Wu, VP of Enterprise AI at IDC. “Runtime security via OpenShell is a foundational step.”

Background
AI agents are increasingly deployed in finance, procurement, supply chain, and manufacturing—where decisions, data, and workflows operate at scale. Huang has described AI as a five-layer stack: energy, chips, infrastructure, models, and applications.
Applications sit at the top, where AI creates economic value. SAP, as a leader in enterprise apps, runs the very workflows where agents must comply with policies, identities, and processes.
What This Means
For enterprises, this collaboration delivers a hardened execution environment for agentic AI in production. Business agents will understand roles, permissions, and data boundaries—limiting what they see, what they can do, and where inference runs.
NVIDIA brings its own perspective as a longtime SAP customer, giving both companies shared context for enterprise-grade governance. The joint development focuses on runtime hardening, policy modeling, identity integration, and auditing hooks.
Related Articles
- From Repository to Roguelike: Your Step-by-Step Guide to Building a Codebase Dungeon with GitHub Copilot CLI
- Trustworthy AI Agents: How NVIDIA and SAP Are Securing Enterprise Automation
- 10 Ways the gcx CLI Transforms Terminal Observability for You and Your AI Agents
- Fonttrio Debuts Open-Source Font Pairing Registry for shadcn/ui Projects
- Mastering GitHub Copilot CLI: Interactive vs Non-Interactive Modes Explained
- Command Line Defies Obsolescence: Terminal Usage Surges in Modern Computing
- 7 Key Principles for Decentralizing Architecture in the Age of AI
- Why I Switched from OneDrive to Ente Photos: A Privacy-Focused Alternative