NVIDIA Networking Software
Accelerate and optimize networking performance.
Overview
A Comprehensive Software Portfolio for Networking Platforms
NVIDIA networking software offers a comprehensive portfolio that optimizes end-to-end networking platforms by maximizing performance and streamlining operations for agentic AI. It accelerates and offloads networking workloads while enhancing end-to-end visibility, enabling faster time to first token and stronger return on investment.
Products
Key Networking Software Offerings
A comprehensive portfolio ensures efficient and continuous operations for gigascale AI factories.
-
DOCA
-
DSX Air
-
NVIDIA Minerva
-
UFM
-
Cumulus Linux
-
Pure SONiC
NVIDIA DOCA Software Platform
NVIDIA DOCA is the unified software platform that unlocks the full potential of NVIDIA BlueField for agentic AI infrastructure. It provides SDKs, libraries, and production-ready microservices for networking, AI-native storage, runtime security, and telemetry at scale. DOCA also powers NVIDIA Sentry, delivering out-of-band, in-silicon AI agent monitoring and policy enforcement as part of NVIDIA Open Agent Safety Platform.
NVIDIA DSX Air
With DSX™ Air, organizations can build full-stack simulations of their data center infrastructure. By providing a cloud-hosted virtual environment, DSX Air enables day-zero operations before a single piece of hardware is unboxed. This includes design, testing, and validation of network provisioning, automation, security policies, and more.
NVIDIA Minerva
NVIDIA Minerva™ is an end-to-end, real-time, AI-native AIOps software platform purpose-built for AI factories. NVIDIA Minerva increases AI factory yield through proactive maintenance, agentic debugging, full-stack correlation, noise reduction, and network observability. NVIDIA Minerva harnesses full-stack telemetry across GPUs, Spectrum-X™ Ethernet, NVLink™, SuperNIC™s, DPUs, schedulers, and workloads. It delivers measurable gains in GPU hours, MTBI, and MTTR. With flexible deployments supporting turnkey or DIY models, NVIDIA Minerva is the AI operations engine for AI factories at any scale.
NVIDIA Unified Fabric Manager (UFM)
The NVIDIA UFM™ platform revolutionizes data center networking management by combining enhanced, real-time network telemetry with AI-powered cyber intelligence and analytics to support scale-out, InfiniBand-connected data centers.
NVIDIA Cumulus Linux
NVIDIA Cumulus™ Linux architecture is the industry’s most innovative, open-network operating system. It was developed under the guiding principles of easy implementation and management, customization, and scalability.
Pure SONiC
NVIDIA offers pure SONiC, a community-developed, open source, Linux-based network operating system that's been hardened in the data centers of some of the largest cloud service providers.
Release Versions
Software and Firmware for NVIDIA Networking Platforms
Next Steps
Ready to Get Started?
Get a Free Trial of NVIDIA Networking Software
Simulate a fully automated network topology using NVIDIA DSX Air.
Discover Networking for the Era of AI
The network is ultimately responsible for AI performance, acting as the backbone of the data center to harness the power of generative AI.
Learn About the Spectrum-X Ethernet Platform
The Spectrum-X Ethernet platform is designed specifically to improve the performance and efficiency of Ethernet-based AI infrastructure.
NVIDIA Networking Software FAQs
NVIDIA offers a comprehensive portfolio of software products to optimize end-to-end networking platforms by maximizing performance and streamlining operations for AI factories. NVIDIA networking software products accelerate and offload workloads while enhancing end-to-end visibility, enabling faster time to first token and stronger return on investment. Products range from network operating systems such as Cumulus and SONiC to fabric management and observability with NVIDIA Minerva and UFM to AI factory simulation with DSX Air.
NVIDIA DOCA is the unified software platform that unlocks the full potential of the NVIDIA BlueField DPU and NVIDIA ConnectX SuperNIC for agentic AI infrastructure. It provides SDKs, libraries, production-ready microservices, and orchestration frameworks to build and operate accelerated infrastructure services—including networking, AI-native storage, runtime security, telemetry, and lifecycle management—at scale.
NVIDIA DSX Air is a logical simulation platform that lets organizations build full-stack digital twins of their data center infrastructure and run validated workloads before hardware is deployed. It supports day-zero operations, including network design, provisioning testing, automation validation, and security policy verification—accelerating time to token for AI factories.
NVIDIA Minerva is an end-to-end, AI-native AIOps software platform purpose-built to maximize the availability and performance of every layer of the AI factory stack. It ingests telemetry from GPUs, Spectrum-X Ethernet and NVLink switches, SuperNICs, DPUs, schedulers, and AI workloads, then applies machine learning and agentic AI to deliver real-time anomaly detection, automated root cause analysis, and closed-loop remediation. In early deployments, NVIDIA Minerva has demonstrated a 20% increase in mean time between interrupts (MTBI), a 63% decrease in mean time to repair (MTTR), up to 99.9% reduction in alert noise, and a 2% increase in effective GPU hours—directly improving token production and AI factory yield.
NVIDIA UFM (Unified Fabric Manager) is a data center networking management platform for InfiniBand scale-out AI clusters. It combines real-time network telemetry with AI-powered cyber intelligence and analytics to manage and optimize NVIDIA Quantum InfiniBand fabrics.
NVIDIA Cumulus Linux is a network operating system (NOS) built on Linux for data center Ethernet networking. It enables complete interoperability, automation-first CI/CD workflows, and flexible open architecture—all manageable with existing Linux tools and talent. Key capabilities include EVPN, virtual routing and forwarding (VRF), and the NVUE CLI for advanced programmability.