Fortinet, Nvidia embed firewall security in AI fabrics

Fortinet has integrated its FortiGate virtual firewall with Nvidia's BlueField-3 data processing unit as part of a joint push into AI data centre infrastructure.

The move embeds firewalling, segmentation and policy enforcement directly on the DPU. It shifts core security processes from the host server CPU into the data centre fabric.

Fortinet said the approach targets private cloud and AI environments that run at high throughput and low latency. It also targets operators that need stronger workload isolation.

John Whittle, Chief Operating Officer at Fortinet, said AI and edge build-outs are straining traditional designs.

"As enterprises are modernizing their data centers to support AI, private cloud and edge applications require much higher throughput than traditional workloads," said John Whittle, Chief Operating Officer, Fortinet. "Integrating FortiGate VM on BlueField-3 DPU gives customers a practical way to keep security aligned with these new performance demands."

The FortiGate VM software now runs directly on the BlueField-3 DPU. This device sits on the server and handles networking and infrastructure services in hardware.

The companies said this design treats the data centre as an "AI factory" with security embedded in the fabric. It moves inspection closer to the network and away from general-purpose compute.

Kevin Deierling, Senior Vice President of Networking at Nvidia, said the trend mirrors other offloaded services.

"AI factories demand an entirely new class of secure, accelerated infrastructure," said Kevin Deierling, Senior Vice President of Networking, NVIDIA. "By running FortiGate VM directly on NVIDIA BlueField-3 DPUs, we're extending the model of infrastructure-offloaded services to include advanced security. This collaboration allows organizations to enforce firewalling, segmentation and zero-trust policies at line rate, without impacting GPU workloads."

Security on the DPU

FortiGate VM is based on the FortiOS operating system. It provides next-generation firewall functions for hybrid cloud and multi-cloud deployments.

In the new configuration, FortiGate runs as a workload on BlueField. The DPU handles firewalling, segmentation and zero-trust policy enforcement.

The host CPU is no longer responsible for these security tasks. It can reserve cycles for AI and other application workloads.

Fortinet said this shift reduces performance impact on AI inference and training jobs. It also improves multitenant isolation and inspection accuracy.

Segmentation policies run within the infrastructure fabric rather than only at the network edge. This gives operators more granular control inside high-density computing clusters.

AI traffic demands

AI and private cloud architectures use large GPU clusters and high-speed interconnects. These clusters drive higher east-west traffic inside the data centre.

Fortinet said the BlueField-based design allows "zero-impact" inspection on the host. The DPU executes the security stack in a separate trust domain from compute workloads.

The companies position this as a better fit for multitenant AI services, edge deployments and service provider environments. These environments often need strict isolation between tenants and services.

Deployment model

Fortinet has produced a deployment guide for the joint solution. The guide describes configuration using standard Open vSwitch bridges.

The design uses OVS switches for wide-area network traffic and VXLAN tunnels inside the data centre. FortiGate VM images run on BlueField-enabled servers.

Fortinet said the approach suits cloud service providers, telecom edge sites and enterprise private clouds. These operators often use service chaining and network overlays at scale.

The companies said customers can use the same Fortinet policy framework across on-premise, cloud and AI factory estates.