Fortinet brings AI security to Equinix distributed hub

Fortinet has made FortiAIGate available on the Equinix Distributed AI Hub, adding a validated reference architecture for securing distributed AI environments.

The arrangement targets organisations moving AI workloads across clouds, data centres and edge environments, where models, data, applications and agents are no longer kept in one place. In that setting, companies face questions around governance, data protection, regional requirements and visibility into how AI tools are used.

In the architecture, FortiAIGate serves as a control point for AI traffic. The companies said it can inspect prompts and responses, apply security policies to AI interactions and provide oversight of usage across environments connected through the Equinix Distributed AI Hub.

Equinix's hub is designed to connect AI agents, models, data, compute resources and cloud providers through private interconnection. The reference architecture places security closer to workloads and data, rather than treating it as a separate layer added later, Fortinet said.

Anthony James, Executive Vice President of Marketing at Fortinet, said, "Enterprise AI is quickly moving beyond isolated applications to distributed environments where models, agents, data, and infrastructure span clouds and geographies. Security needs to be built into that architecture from the start. By bringing FortiAIGate to the Equinix Distributed AI Hub, we are giving customers a simple way to secure AI interactions, protect sensitive data, maintain control over where that data is processed, and scale AI with confidence."

The companies cast the development as part of the shift from AI experimentation to production deployment. As businesses expand AI use, they must manage not only model access and spending, but also the risk of sensitive information moving into external systems without adequate controls.

Governance focus

Within the reference design, FortiAIGate applies policies consistently across approved large language model traffic and model context protocol traffic. Customers can also track per-token spending and set budgets, limits and alerts, Fortinet said, reflecting how cost management is becoming part of AI governance rather than a separate procurement issue.

The companies also highlighted data residency and sovereignty. Control, management and inspection can remain within customer infrastructure, according to Fortinet, which said this would help organisations meet region-specific governance requirements or keep sensitive data within defined jurisdictions.

Another part of the pitch is visibility into sanctioned and unsanctioned AI use. Fortinet said FortiAIGate is intended to govern approved AI interactions, while FortiGate can provide visibility into unauthorised, or shadow, AI activity across the network.

Maryam Zand, Vice President of Partnerships & Ecosystem Development at Equinix, said, "AI isn't centralised, which means enterprises need security that extends across every environment. Enterprises need to connect models, data, and clouds while retaining control over their architecture and security choices. Our distributed AI hub brings those resources together through private interconnection. The validated architecture with Fortinet gives customers a way to apply FortiAIGate security and governance to AI interactions across that connected environment."

Wider tie-up

The announcement also reflects a broader relationship between the two companies. Fortinet and Equinix already work together on network security through FortiGate VM on Equinix Network Edge, and the new reference architecture extends that collaboration into AI traffic inspection and governance.

That matters because many enterprises are trying to stitch together AI systems spanning internal infrastructure, third-party cloud services and external models. Security suppliers are increasingly seeking to position themselves not just around endpoint or network protection, but around the flow of prompts, data and decisions between AI components.

Fortinet said the architecture forms part of its AI-Native Security Fabric, linking AI security, networking and infrastructure protection. In practical terms, the joint design is intended to give customers a single approach to protecting network connectivity, data movement and interactions among applications, models, agents and users.

The companies also said the design can use graphics processing units, including those from NVIDIA, to handle inspection at the network layer without affecting AI workloads. That addresses a concern among organisations adopting AI tools at scale, where additional security checks can add delay or constrain throughput.

The announcement underlines how infrastructure providers and cybersecurity companies are trying to define guardrails for enterprise AI as deployments become more distributed.