Fortinet is expanding FortiAIGate with Nvidia platforms and software to secure AI workloads, data, and AI agents. The integration is designed to allow organizations to set up zero-trust AI environments with GPU-accelerated security in data centers, the cloud, and at the network edge.
Fortinet announces that FortiAIGate will work more closely with various Nvidia technologies. This involves an integration of the security solution with Nvidia Blackwell, Nvidia Hopper, Nvidia Dynamo, and Nvidia Nemotron security models.
The solution sits inline between applications and AI models. This allows FortiAIGate to monitor traffic, check model input and output, and log suspicious interactions without organizations having to add separate security layers.
Security tailored to the GPU
The expansion aims to accelerate AI security via GPUs instead of just CPUs. According to Fortinet, this allows runtime guardrails to be applied to large language models, MCP servers, and AI agents, with limited latency and higher throughput.
FortiAIGate uses Nvidia Dynamo, an open-source framework for distributed inference, among other things. The combination with Blackwell and Hopper GPUs is intended to ensure that security takes less computing power away from other business workloads.
Zero trust
FortiAIGate is also designed to extend zero-trust principles to AI environments. The solution manages all AI model data traffic, blocks prompt injection attacks, and filters unwanted or dangerous output to ensure sensitive data remains protected.
Additionally, Fortinet emphasizes self-hosting and local control. Organizations can build and manage AI solutions on their own infrastructure, with monitoring of user and device interactions and logging of suspicious prompts and responses, which should help with privacy regulations such as GDPR.
From data center to edge
FortiAIGate will be available as a GPU-supported appliance for data centers, as well as a virtual appliance or container on Nvidia-certified systems. The solution thus supports on-premises, cloud, edge, and hybrid deployments from central management.
For larger environments, Fortinet is also targeting multi-tenant use. Through Nvidia’s multi-instance GPU virtualization, workloads and datasets from different customers can remain isolated from each other on the same hardware, with guarantees regarding performance and fault isolation.