NVIDIA Solutions from Arrow

NVIDIA AI Enterprise

NVIDIA AI Enterprise is a comprehensive software suite that provides a consistent and optimized AI development and deployment experience across NVIDIA’s hardware platforms, including IGX and Jetson. By leveraging containerized microservices, NVIDIA AI Enterprise empowers you to optimize resource allocation for specific workloads.

Talk to an Expert

The End-to-End Software Platform for Production AI

NVIDIA AI Enterprise provides leading development tools, frameworks, and pre-trained models, alongside robust management and orchestration capabilities, ensuring optimal performance, high availability, and security.

The NVIDIA AI Enterprise provides a range of tools and frameworks specifically designed to streamline the process of building and deploying AI applications at the edge. Notable tools include:

NVIDIA Inference Microservices (NIMs): A set of easy-to-use microservices designed to accelerate deployment of generative AI models across cloud, data center, and workstations. Supporting a wide range of AI models, including NVIDIA AI Foundation and custom models, it ensures seamless, scalable AI inferencing.
NVIDIA CUDA-X: Built on top of CUDA®, is a collection of microservices, libraries, tools, and technologies for building applications that deliver dramatically higher performance than alternatives across data processing, AI, and high-performance computing (HPC).
Visual Insight Agent (VIA): A collection of workflows to build AI agents capable of processing large amounts of live or archived videos and images with Vision-Language Models (VLM). VIAs will help nearly every industry summarize, search, and extract actionable insights from video using natural language.
NVIDIA cuOpt: A microservice to accelerate logistics and supply chain optimization to save time and reduce infrastructure costs.

NVIDIA Omniverse

NVIDIA Omniverse is a platform for building 3D applications and simulations. It uses real-time technologies and combines them with Universal Scene Description (USD), a format enabling seamless collaboration across different 3D tools. This allows developers and creators to connect and work together on massive, physically accurate virtual worlds.

Contact Our Experts

Platform Components

The Omniverse platform consists of the following components:

Omniverse Kit SDK is a powerful toolkit for developers to easily build advanced, custom OpenUSD tools and plugins quickly that are interoperable across 3D design tools
Developer Tools to Build, Test, Launch, and Package applications
Omniverse Nucleus is the database and collaboration engine, allowing users to store, share, and collaborate on OpenUSD project data
Omniverse Simulation provides realistic physics simulation for rigid and soft bodies, smoke, fire, water, destruction, and fracture, powered by NVIDIA PhysX
Omniverse RTX Renderer is an advanced, AI-enabled renderer based on NVIDIA RTX that supports real-time ray tracing, interactive path tracing, and accurate NVIDIA Iray rendering

Datacenter GPUs (Passive Cooling)

Built for raw power and handling massive workloads, the NVIDIA Ampere architecture is ideal for complex simulations, artificial intelligence (AI), and rendering huge datasets. They often come with features like high memory capacity and specialized software for virtualization, allowing them to serve multiple users remotely.

Explore Datacenter GPUs

Workstation GPUs (Passive and Active Cooling)

NVIDIA RTX GPUs excel in tasks like video editing, 3D animation, and running demanding design software. They focus on smooth graphics rendering and responsiveness, making them ideal for creative professionals and engineers working on their own machines.

Explore Workstation GPUs

DGX AI Compute Systems

NVIDIA DGX are AI computer systems specifically designed to handle the demanding needs of artificial intelligence. They are powerhouses built for data scientists and researchers working on large-scale projects.

The latest DGX systems are powered by NVIDIA’s Blackwell architecture. This cutting-edge technology boasts significant performance improvements over previous generations. Specifically, Blackwell offers up to 15x faster real-time processing for complex AI models with trillions of parameters. This translates to benefits in areas like generative AI, a field that’s powering advancements in everything from drug discovery to creating realistic simulations.

Contact Our Experts

NVIDIA Networking Solutions

NVIDIA offers a comprehensive suite of networking solutions that enable high-performance, scalable, and feature-rich connectivity for a wide range of applications, from traditional enterprise needs to the most demanding AI, high-performance computing, and hyperscale data center workloads.

NVIDIA’s networking portfolio includes:

Spectrum Ethernet Switch portfolio that provides up to 800GbE low-latency interconnectivity within data centers
ConnectX network adapters that support both InfiniBand and Ethernet protocols, delivering low-latency and high-performance services

Talk to an Expert

Spectrum Ethernet Switch Portfolio

Details

Up to 800Gb/s per port for 51.2Tb/s aggregate switch bandwidth
SN5400: Smart-leaf, spine, and super-spine switch offers 64 ports of 400 GbE in a 2U form factor
SN5600: Smart-leaf, spine, and super-spine switch offers 64 ports of 800 GbE in a dense 2U form factor

NVIDIA Spectrum-4 SN5000 Series

Details

Provides port speeds spanning from 1 to 400 gigabits per second (Gb/s)
SN4700: Spine/super-spine switch offers 32 ports of 400GbE in a compact 1U form factor
SN4600: A 2U 64-port 200GbE spine switch used as a high-density leaf, fully splittable to up to 128 10/25/50GbE ports
SN4600C: A 64-port 100GbE switch system that’s ideal for spine/super- spine applications

NVIDIA Spectrum-3 SN4000 Series

Details

Provides port speeds spanning from 1 to 200 gigabits per second (Gb/s)
SN3700: Spine/super-spine switch offers 32 ports of 200GbE (50G PAM4 per lane) in a compact 1U form factor
SN3700C: A 1U 32-port 100GbE spine switch that can also be used as a high-density 10/25GbE leaf
SN3750-SX: A 1U 32-port 200GbE switch that’s supports port speeds spanning 1–200Gb/s per port
SN3420: Switch offers 48 ports of 10/25GbE and 12 ports of up to 100GbE and utilizing 25G NRZ per lane in a compact 1U form factor

NVIDIA Spectrum-2 SN3000 Series

Details

Provides port speeds spanning from 1 to 100 gigabits per second (Gb/s)
SN2201: Management switch or as a ToR switch connecting up to 48 1G Base-T host ports with non-blocking 100GbE spine uplinks
SN2100: Accommodates 16 ports running at 100Gb/s, with a throughput of 1.6 terabits per second (Tb/s) and a 2.38 billion packets per second (Bpps) processing capacity
SN2010: Deliver up to 850Gb/s aggregate bandwidth with 1.26Bpps processing capacity in a compact half-width 1RU form factor

NVIDIA Spectrum SN2000 Series

NVIDIA ConnectX SmartNIC Portfolio

Details

25Gb/s Optimized
Max. Ports and Max. Speeds are 2 x 25Gbps or
1 x 50Gbps NRZ
Up to 50Gb/s total bandwidth

ConnectX-6 Lx Enterprise-Class SmartNIC

Details

Security and Performance
Max. Ports and Max. Speeds are 2 x 100Gbps or 1 x 200Gbps NRZ/PAM4(56)
Up to 200Gb/s total bandwidth

ConnectX-6 Dx Cloud, SDN/NFV SmartNIC

Details

400Gb/s Data Centric Acceleration
Max. Ports and Max. Speeds are 1 x 400Gbps or 2 x 200Gbps NRZ/PAM4(56/112)
Up to 400Gb/s total bandwidth

Accelerate AI Development with Arrow and NVIDIA

Build Cutting-Edge Artificial Intelligence (AI) and High-Performance Computing Solutions