Traefik Labs, the creator of one of the world's leading cloud-native application proxies with over 3.4 billion downloads, has announced significant expansions to its enterprise AI infrastructure platform at Oracle AI World. These additions are designed to deliver AI sovereignty for regulated industries and security-conscious enterprises by eliminating dual lock-in (dependency on specific LLM providers and cloud platforms) and ensuring governance works everywhere.
New Components: Traefik launched the MCP Gateway for AI agent governance and enhanced the AI Gateway with native support for NVIDIA Safety NIMs.
Offline Capability: The entire Traefik platform now supports comprehensive offline deployment for air-gapped environments.
AI Sovereignty: The platform delivers consistent AI governance as infrastructure, allowing organizations to maintain data sovereignty and switch providers easily.
NVIDIA Integration: The AI Gateway natively supports three NVIDIA Safety NIMs (Topic Control, Content Safety, and Jailbreak Detection), which run entirely offline.
Agent Security: The MCP Gateway enforces identity-driven, granular control over AI agents accessing enterprise systems to prevent data exfiltration.
Oracle Integration: These capabilities are available today through Traefik's integration with Oracle Cloud Infrastructure (OCI).
The Traefik AI Gateway has been updated to natively support three essential NVIDIA Safety NIMs: Topic Control, Content Safety, and Jailbreak Detection. These capabilities are chained together to create multi-layered safety pipelines that are critical for enterprise-grade AI protection. Crucially, these pipelines are designed to run entirely offline within the customer's infrastructure. This provides AI safety and governance without requiring organizations to send sensitive data to the public cloud, thus ensuring zero-egress security and complete data sovereignty.
Addressing the growing need for security around autonomous systems, Traefik introduced the new MCP Gateway (Model Context Protocol Gateway). This component provides identity-driven governance specifically for AI agents that access enterprise systems, databases, and APIs. Unlike traditional API gateways that lack agent visibility, the MCP Gateway enforces granular control over the agent's tools, tasks, and transactions. This level of security is achieved by scoping access based on resource type (e.g., database table), operation (read vs. write), and transaction attributes, effectively preventing a compromised or manipulated agent from exfiltrating data or executing unauthorized actions.
A major feature announced is Platform-Wide Offline Deployment. The entire Traefik platform, including the AI Gateway, MCP Gateway, and API Management components, now operates in a comprehensive offline mode across any infrastructure. This means customers can deploy the same solution with identical features and performance from Oracle Cloud and private datacenters all the way to completely air-gapped military installations, with zero external dependencies.
Sudeep Goswami, CEO at Traefik Labs, emphasized the importance of this capability: "Enterprises need infrastructure that's as portable as their applications. We're delivering AI governance that works identically everywhere, from multi-cloud deployments to completely disconnected environments. That's what unblocks enterprise AI."
Industry analysts recognize the importance of these security layers. Paul Nashawaty, Practice Lead and Principal Analyst at theCUBE Research, stated that Traefik is "shifting this dynamic" by offering a prescriptive, modular framework for safety controls.
Ikenna Nwaiwu, Principal Consultant and author, also highlighted the unique value of the new agent controls: "MCP governance can't be addressed with traditional API gateways - the protocol is fundamentally different. What Traefik has built with the MCP Gateway is a real solution for enforcing tool, task, and transaction-level policies on AI agents. This level of granularity is essential for production deployments. Having a single platform that works identically across any environment, including air-gapped facilities, is a massive win for enterprise architecture teams."
These new features are available today through Traefik's integration with Oracle Cloud Infrastructure (OCI). This collaboration extends Oracle's deployment flexibility (across public cloud, Cloud@Customer, and specialized environments like Roving Edge devices) to AI workloads. Organizations can now deploy AI with NVIDIA Safety NIMs and agent governance across Oracle's entire spectrum of deployment options while maintaining crucial operational control and regulatory compliance.
Traefik Labs empowers organizations to adopt and scale cloud-native architectures through its modern unified platform for application connectivity and API management. Traefik Proxy, the company's flagship open-source project, is trusted by the world's largest enterprises and ranks among Docker Hub's top projects, with over 3.4 billion downloads and more than 57,000 stars on GitHub. Founded in 2016, Traefik Labs is backed by prominent European investors including Balderton Capital and Elaia.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.