The NVIDIA Jetson Expansion Card adds up to 100 TOPS of AI inference directly to your Nodegrid Net Services Router — no separate appliance, no additional rack space, no physical access required. Deploy, flash, and manage AI workloads remotely through the same Gen 3 Out-of-Band infrastructure you already rely on for your network. From AI-powered NetOps and agentic network automation to real-time video analytics and sensor processing, the card transforms your NSR into an intelligent edge platform that can observe, decide, and act locally.

• Reduce CAPEX & OPEX – Consolidate AI compute, networking, and OOB management into a single platform instead of deploying and maintaining standalone edge AI appliances at every site.
• Reduce downtime & truck rolls – Remotely flash JetPack, OS, and BIOS, access the Jetson console, and power cycle or recover the module over OOB — capabilities that traditionally require a technician with a USB cable on-site.
• Minimize MTTR & operational risk — Centralized remote access and control through Nodegrid OS means faster diagnosis, faster recovery, and fewer variables when troubleshooting edge AI workloads.
• Increase site reliability — Purpose-built to work within the proven NSR platform using open, industry-standard hardware and the same intuitive Nodegrid OS interface your team already knows.

AI for the Edge

• Run real-time AI inference for AI-powered NetOps, agentic network automation, video analytics, and predictive maintenance — with up to 100 TOPS of compute from NVIDIA Jetson Orin modules
• Enable agentic network operations where the edge can autonomously detect anomalies, correlate events, and trigger remediation — without waiting for centralized decision-making
• Deploy and manage AI workloads remotely through Nodegrid’s Gen 3 OOB infrastructure — no physical access, no USB cables, no truck rolls
• Process video, sensor (GPIO/I2C), and telemetry data locally to reduce cloud dependency and latency
• Consolidate AI compute, networking, and OOB management into a single, compact expansion card that fits any NSR 1U or NSR 2U slot

Avoid Truck Rolls — Full Remote OOB for Jetson

Standard NVIDIA Jetson deployments require physical access for critical operations: flashing JetPack or OS updates demands a direct USB cable connection to a host workstation, and the device must be manually placed into Force Recovery Mode. Console access similarly requires a physical serial or USB-C cable.

The NVIDIA Jetson Expansion Card eliminates these constraints. Through the NSR’s Gen 3 Out-of-Band management, IT teams can remotely:

• Flash JetPack, OS, BIOS, and AI application images — no USB cable, no host workstation on-site
• Access the Jetson serial console over the network for troubleshooting and configuration
• Power on, power off, reset, and invoke recovery mode — all without touching the device

Connect your Jetson AI applications directly to the NSR through a dedicated 1GbE interface for reliable edge computing and network automation.

This is the industry’s first and only solution that extends Out-of-Band management to NVIDIA Jetson edge AI modules, enabling true lights-out AI deployment at scale.

Up to 100 TOPS AI Inference

The expansion card supports two NVIDIA Jetson Orin module options to match workload requirements:

• Orin Nano 8GB — Up to 67 TOPS of inference performance for AI NetOps tasks such as network telemetry analysis, event correlation, sensor analytics, and telemetry preprocessing.
• Orin NX 8GB / 16GB — Up to 100 TOPS of inference performance for demanding workloads including agentic network automation, multi-stream video analytics, object detection, and hardware-accelerated video encoding and decoding.

Both modules are purpose-built for continuous, always-on inference at the edge — delivering the compute necessary for autonomous network operations, real-time decision-making, and local action without backhauling data to the cloud.

Versatile Edge Compute — Video, Sensor, and GPIO

The NVIDIA Jetson Expansion Card is not limited to a single workload type. Its onboard I/O enables a wide range of AI and non-AI applications at the edge:

• Video — 1x HDMI output and hardware-accelerated encoding/decoding (NX) for real-time video analytics, NVR workloads, and quality inspection
• Sensor and control — 2x GPIO and I2C (SDA/SCL) interfaces for direct connection to environmental sensors, actuators, and industrial control devices
• Connectivity — 2x USB 3.0 ports and 1x console port for peripheral expansion, external storage, or additional device interfaces

Whether the use case is agentic network operations at a large branch, video analytics at a co-location facility, sensor aggregation at a remote utility site, or network-driven automation at a cell tower, the card adapts to the environment.

Technical Specifications