DRACO Jetson Thor Rugged Edge AI Computer

Overview

Datacenter-class AI silicon in a fanless box built for a vehicle. NVIDIA Jetson Thor delivers 2070 FP4 TFLOPS against 128 GB of unified memory at 273 GB/s — and this chassis surrounds it with what edge deployments actually need: 8× GMSL2 camera inputs, 5× M12 10GbE with PTP time sync, 4× CAN, centimetre-grade RTK positioning and a 10–36 V input with ignition sensing.

Key highlights

• NVIDIA Jetson Thor T5000 — 2070 FP4 TFLOPS.
• 128 GB unified LPDDR5X · 273 GB/s — CPU and GPU share one memory pool.
• Fanless construction — no moving parts to fail on site.
• Full CUDA and JetPack — the same toolchain as every larger NVIDIA system RDP supplies.
• Production-sealed enclosure using the module you prototyped on.
• Rugged: wide-range DC with ignition sensing, sealed against dust and vibration.
• Imaged and validated by RDP against your application before dispatch; GST invoice, pan-India onsite, GeM.

AI workload fit

• Multi-camera computer vision — inspection, safety, surveillance.
• Real-time inference where latency or connectivity rules out the cloud.
• Agentic AI running unattended at a remote site.
• NLP & speech processing of local audio.
• Generative AI at the edge for on-site content and reporting.
• Volume rollout across sites, vehicles or plant.

AI workload positioning

Edge deployments fail on interfaces and environment far more often than on compute. Camera link type, time sync, power tolerance and ingress protection decide whether a unit survives the site; the accelerator rarely does. Specify those with the deployment, not after it.

Industry use cases

• Manufacturing: line inspection, safety zones and predictive maintenance.
• Public sector & sovereign: smart-city, traffic and rail-transit analytics, on soil.
• Telecom & 5G: inference at the network edge and in cell-site cabinets.
• Automotive & mobility: AMR/AGV fleets and off-highway machinery.
• Defence & aerospace: inspection drones and rugged field systems.

Performance & how to be sure

Precision basis matters enormously at the edge — an FP4 TFLOPS figure is not comparable with an INT8 TOPS number, and vendors quote whichever flatters. RDP will run your actual pipeline on the unit rather than restate a headline. NVIDIA publishes 2070 FP4 TFLOPS for this module; any “Super mode” figure is a power/clock mode rather than the default and carries a real thermal cost. Rather than quote a tokens/sec figure that will not match your workload, RDP offers a “benchmark your model” session: bring the model, precision and context length you actually intend to run, and we will validate it on this exact configuration before you commit.

Series & upgrade path

DRACO tier. The Jetson ladder runs Orin Nano → Orin NX → AGX Orin → Thor; step up when model size or camera count exceeds the tier, not by default. Above the ladder: RDP micro data center pods when a site outgrows a single node.

On-prem vs cloud (TCO)

For sustained daily AI work this configuration removes per-GPU-hour billing, queueing for scarce instances, and egress charges on your own data — and keeps everything on-premises for DPDP and data-residency obligations. Cloud still wins for burst capacity and one-off very large training runs; on-prem wins on sustained utilisation, control and predictable INR capital cost. We model the crossover with your actual usage rather than assert it.

Software & day-one readiness

Ships with NVIDIA JetPack — the Jetson SDK — including CUDA, cuDNN, TensorRT, DeepStream for multi-camera video analytics, and support for the Isaac (robotics) and Metropolis (vision) frameworks. This is the same CUDA programming model as every larger NVIDIA system RDP supplies, so a model developed on a workstation or GPU server deploys to the edge without a rewrite. RDP images and validates the unit against your application before dispatch.

Power, thermal & acoustics

Wide-range DC input with ignition sensing and over/under-voltage protection. Fanless — no moving parts to fail in dust, vibration or humidity. Enclosure rating, mounting and operating-temperature range confirmed per deployment rather than assumed. Exact wattage, BTU and dB(A) figures come from RDP bench measurement rather than estimates — ask for the site-readiness sheet with your quote.

Deployment, warranty & support

Built to order by RDP Technologies. Supplied with GST invoice, pan-India onsite support, and availability through GeM for public-sector procurement. Built to order — typically 8–10 weeks; module allocation confirmed at quotation.

Why RDP

RDP Technologies is a Make-in-India OEM with 14+ years and 300,000+ units shipped, supplying AI infrastructure from desk-side systems to rack-scale AI factories — with predictable INR pricing, GST invoicing and pan-India onsite service.

Buy with confidence

Use Request a Quote to reach an RDP solution architect for sizing, a benchmark-your-model session, financing options and a delivery plan. No obligation.

Specifications