Fortinet expands NVIDIA tie-up to secure enterprise AI
Wed, 13th May 2026 (Today)
Fortinet has expanded its FortiAIGate integration with NVIDIA, focusing on securing enterprise artificial intelligence deployments across data centres and cloud environments.
FortiAIGate sits between applications and AI models, monitoring traffic and applying controls to prompts and outputs. The integration uses NVIDIA AI platforms and software to protect AI workloads, data and autonomous agents in real time.
The companies are positioning the product as businesses move large language models and AI agents into production while trying to limit data leakage, misuse and governance failures. It is aimed at organisations running AI on their own infrastructure, in the cloud, or across hybrid and edge environments.
The system supports self-hosted deployments, helping organisations keep AI development and inference within their own infrastructure. This is targeted at companies and public sector bodies facing data sovereignty rules or privacy requirements linked to where data is stored and processed.
Security focus
A central part of the pitch is the application of zero-trust principles to AI systems. FortiAIGate manages traffic to AI models, blocks prompt injection attempts and filters toxic or unauthorised content, while logging prompts and responses linked to suspicious incidents.
The product also extends protection beyond models to related systems such as MCP servers and AI agents. The aim is to give security teams more direct oversight of how AI tools are used inside an organisation.
Fortinet is also emphasising the operational case for using graphics processors rather than conventional CPU-based infrastructure for this type of security layer. The system runs on NVIDIA Blackwell GPUs and NVIDIA Hopper, and uses the NVIDIA Dynamo inference-serving framework.
The argument is that AI security controls need to operate inline without creating delays that make AI services harder to use. The arrangement is intended to deliver low-latency inspection while reducing hardware footprint, server load and energy use.
Deployment options
Deployment flexibility is another part of the announcement. FortiAIGate can be used as a GPU-based appliance in data centres, as a virtual appliance, or as containers on NVIDIA-Certified Systems.
That gives customers a range of options depending on whether they run AI services on premises, in public cloud environments, or in mixed settings. Centralised management remains available across those models.
Fortinet also highlighted support for multitenant environments, which are common in large enterprises and AI data centres. NVIDIA virtualisation methods can partition resources and isolate workloads or customer datasets on the same hardware.
Among the technologies referenced was NVIDIA Multi-Instance GPU, which allows a single physical GPU to be divided into separate instances. This is designed to prevent one AI service from affecting the performance or stability of another.
Executive comments
The announcement included comments from both companies on the commercial demand behind the tie-up.
"Enterprises everywhere are racing to adopt AI, and security has become a critical enabler of that innovation. Together with NVIDIA, we're delivering a solution that helps organizations secure and optimize AI deployments while maintaining performance, controlling costs, and meeting data sovereignty requirements. FortiAIGate combines Fortinet's AI-driven Security Fabric with NVIDIA's high-performance computing and AI factories to stop threats, from malicious prompts to data exfiltration, without disrupting AI workflows," said John Whittle, Chief Operating Officer, Fortinet.
NVIDIA linked the partnership to the growth of autonomous AI systems inside businesses.
"The accelerating shift toward autonomous AI agents is creating unprecedented demand for secure, high-performance enterprise computing platforms. By integrating its FortiAIGate solution with the full-stack NVIDIA AI platform, Fortinet provides zero-trust security and real-time governance, reducing threat exposure by shortening response times," said Justin Boitano, Vice President, Enterprise AI Platforms, NVIDIA.
The announcement reflects a broader industry effort to build security controls around generative AI and agent-based systems as they move from trials into production use. Vendors across cybersecurity, infrastructure and cloud computing are trying to establish themselves in a market where customers want tighter control over model access, user behaviour and sensitive data flows.
FortiAIGate is intended to support that shift by giving organisations a way to govern AI systems across different deployment models while keeping inspection close to the model and application layer.