Fortinet unveils Secure AI Data Centre with 69% power saving

Fortinet has introduced its Secure AI Data Centre solution, which is designed to provide security across the entire artificial intelligence (AI) stack, from physical infrastructure to applications and large language models.

The framework integrates advanced threat protection measures and claims average power consumption reductions of 69 per cent compared to conventional security approaches.

Comprehensive framework

The Secure AI Data Centre solution aims to secure AI workflows with features spanning network segmentation, encrypted traffic inspection, application-level controls, and defences against data and prompt injection. Fortinet says the protection model supports local, hybrid, and public cloud deployments, while enforcing industry compliance standards for AI use within data centres.

Through embedded security at each layer, the framework is intended to prevent data leakage, stop malicious inputs, and help enterprises manage compliance requirements within high-density GPU environments. A consolidated management plane, the Fortinet Security Fabric, unifies firewall, application security, and AI protections to reduce audit time and operational complexity.

New firewall unveiled

Central to the solution is the introduction of the FortiGate 3800G firewall, engineered for data centres running intensive AI workloads. The device is powered by NP7 and SP5 application-specific integrated circuits (ASICs) and is equipped with 400 GbE connectivity. According to Fortinet, this enables hyperscale throughput with energy efficiency for GPU clusters, addressing the demand for scalable AI security with a focus on operational cost and power savings.

The FortiGate 3800G builds on Fortinet's existing data centre firewall portfolio, targeting real-time protection for large-scale GPU clusters and data pipelines. According to the company, this hardware is intended to shield AI workloads while supporting sustainability goals by lowering energy cost per gigabit per second processed.

Layered AI defence

Other features highlighted include AI model and API security, which implement layered controls to inspect and clean all traffic before it reaches the AI models, aiming to prevent code injection, tampering, or data theft. The framework also incorporates quantum-resistant cryptography and quantum key distribution technologies to prepare for future threats posed by quantum computing advances.

The framework's controls for large language models help protect against prompt injection, data leakage, and misuse through the management of model traffic and the enforcement of guardrails on inputs and outputs.

User feedback

"AI data centres demand both massive performance and deep protection. Our Secure AI Data Centre solution unifies those capabilities, combining ASIC-powered firewalls like the FortiGate 3800G with advanced protection for data, applications and LLMs so organisations can scale AI without compromising security, performance, or efficiency," said Nirav Shah, Senior Vice President, Products and Solutions, Fortinet

Huy Ly, Head of Global IT Security at Monolithic Power Systems, described the deployment of the solution in a production environment. "AI is transforming every aspect of our business, from product design to supply chain management, while introducing new operational challenges. The Fortinet Secure AI Data Centre solution gives us the visibility, performance, and protection we need to operate high-density GPU clusters with confidence. With this solution integrated into our AI environment, we can safeguard sensitive models and data while maintaining hyperscale throughput with greater efficiency and cost performance," said Ly.