Fortinet is going deeper with Nvidia, announcing an expanded integration of its FortiAIGate solution with Nvidia’s AI platforms. The partnership, revealed on May 12, aims to deliver GPU-accelerated, zero-trust security built specifically for AI workloads running in data centers and cloud environments.
The integration taps Nvidia’s Blackwell and Hopper GPUs alongside the Dynamo inference framework to protect against AI-specific threats, including prompt injection attacks and data exfiltration, while keeping latency to a minimum.
What the integration actually does
Fortinet’s approach embeds security controls directly within the AI compute layer, so threats targeting large language models and agentic AI systems get caught at the point of execution. When an enterprise runs AI models on Nvidia hardware, Fortinet’s security layer operates at GPU speed alongside those models, inspecting and enforcing policies without creating bottlenecks.
The partnership specifically addresses runtime guardrails for LLMs and autonomous AI agents. Fortinet’s solution also supports scalable multitenant deployments, which matters for cloud providers and managed service companies running AI workloads for multiple clients on shared infrastructure. Data sovereignty is another focal point of the integration.
