Cisco and NVIDIA have announced a new suite of security and performance solutions designed to set a higher benchmark for enterprise artificial intelligence environments. Revealed at the Cisco Live conference in San Diego, the new Cisco AI Defense and Hypershield offerings integrate NVIDIA´s artificial intelligence technologies to provide full visibility, validation, and runtime protection for artificial intelligence workflows throughout their lifecycle. This move extends the Cisco Secure AI Factory framework, originally launched with NVIDIA during NVIDIA´s GTC conference in March, emphasizing the importance of trust and security as artificial intelligence becomes foundational across industries.
The Cisco Secure AI Factory with NVIDIA delivers continuous monitoring and security from data ingestion through model training to inference and deployment. Cisco AI Defense and Hypershield are engineered to operate in tandem with NVIDIA artificial intelligence solutions, enhancing scalability, performance, and trustworthiness for complex agentic and generative workloads. NVIDIA´s Enterprise AI Factory validated design now incorporates these Cisco tools, enabling organizations to protect every phase of their artificial intelligence deployments and confidently scale operations.
Open models that have been post-trained using NVIDIA NeMo and secured with NVIDIA Blueprints can be validated through Cisco AI Defense, bolstered by microservices such as NVIDIA NIM for optimized inference in production environments. Cisco´s AI Defense provides granular runtime monitoring for artificial intelligence agents on the NVIDIA AI platform, while Cisco Hypershield will soon integrate with NVIDIA BlueField DPUs and the DOCA Argus framework to extend real-time, distributed security and threat detection to every node in the infrastructure. The joint solutions allow organizations to adopt zero-trust security across both data center and edge environments, maintaining protection regardless of data or workload location.
Network performance has also been addressed, with Cisco and NVIDIA introducing intelligent packet flow management, NVIDIA Spectrum-X for AI-optimized Ethernet connectivity, and end-to-end visibility spanning networks and GPUs. These enhancements allow for proactive issue detection, vital for data-hungry and latency-sensitive artificial intelligence tasks. Cisco is further expanding its modular AI PODs—scalable, validated blocks designed for training, fine-tuning, and inference—giving organizations greater flexibility and control as they scale artificial intelligence initiatives. Additionally, the new NVIDIA RTX PRO 6000 Blackwell Server GPU is now available for Cisco UCS C845A M8 servers, promising faster and more powerful processing for the most demanding artificial intelligence applications.
