Sunnyvale, United States - December 17, 2025 - Fortinet and Arista Networks have jointly announced a Secure AI Data Center solution, a validated reference architecture that combines best-of-breed networking and cybersecurity to help enterprises deploy and scale AI infrastructure with both high performance and zero-trust protection from the core network to model inference workloads.
The collaboration builds on Fortinet’s Secure AI Data Center framework by integrating Arista’s ultra-low-latency, high-performance networking technology with Fortinet’s ASIC-accelerated firewalls, AI-aware threat protection, zero-trust segmentation, encrypted traffic inspection, and automated response capabilities.
Together, the vendors say the architecture offers a modular, scalable approach to securing large GPU clusters, data pipelines, and LLM workloads without introducing unnecessary operational complexity.
“AI infrastructure is advancing faster than most organisations can secure or operationalise it,” said Nirav Shah, Senior Vice President of Products and Solutions at Fortinet. “By collaborating with industry leader Arista Networks, we’re delivering a secure AI data center architecture and integrated solution that brings security, performance, and operational simplicity together from day one. This alliance gives customers a validated blueprint to scale AI with confidence, without compromising protection or creating unnecessary complexity.”
The joint solution targets core barriers to AI adoption that organisations increasingly face, including infrastructure costs, vendor lock-in, performance bottlenecks, and security threats ranging from model tampering to data leakage.
The partners say the architecture delivers zero-trust protection from the network core through model inference, hardware-accelerated security that enhances AI performance, an open modular design to avoid lock-in, and automated, self-healing capabilities for real-time detection and remediation.
At the heart of the design, Fortinet’s ASIC accelerates HTTPS/TLS offload to deliver up to 33× performance with sub-single-microsecond latency, freeing server CPUs and memory for forward AI workloads and reducing networking contention, jitter, and tail latencies under high load. Arista’s networking gear contributes ultra-low-latency switching and cluster load balancing, addressing the throughput and connectivity demands of hyperscale AI/ML clusters.
Ed Chapman, Vice President of Business Development at Arista Networks, highlighted the strategic fit, “Our collaboration with Fortinet brings together best-in-class cybersecurity with Arista’s ultra-low-latency, high-speed networking, which is essential for AI applications.”
The partners are already validating the architecture in real-world environments with Monolithic Power Systems (MPS), where the integrated solution is being deployed to secure high-density GPU clusters and deliver consistent security and performance at scale.
MPS’s Huy Ly, Head of Global Infrastructure/Security, said the reference architecture gives the organisation “the visibility, performance, and protection we need to operate high-density GPU clusters with confidence,” while maintaining hyperscale throughput and operational efficiency.
By converging advanced networking, security, and AI infrastructure components, the Secure AI Data Center solution aims to reduce operational risk, streamline operations, and improve compute ROI for enterprises rolling out modern AI workloads.
