900-13767-0000-0A1 - Jetson Orin NX 16GB TAA Module | NVIDIA
MPN： 900-13767-0000-0A1 ✓ 啟用中900-13767-0000-0A1 概覽
A System-on-Module (SOM) is a complete embedded computing subsystem that integrates a CPU, GPU, memory, power management, and connectivity onto a single board that plugs into a carrier board. SOMs belong to the embedded computing taxonomy (SoM -> embedded module -> SBC -> edge computing node) and accelerate time-to-market by abstracting high-speed memory routing, PMIC design, and thermal engineering from the customer's product team.
Key features include 100 TOPS of INT8 AI inference, 2x NVIDIA NVDEC for hardware video decode, 1.4x NVENC for hardware video encode, dual MIPI CSI-2 camera interfaces, PCIe Gen 4.0 lanes, and support for the JetPack SDK with CUDA, TensorRT, and DeepStream SDKs. The module draws 10 W to 25 W configurable power, making it suitable for fan-cooled edge AI deployments where thermal envelope is constrained.
The Jetson Orin NX 16GB uses the NVIDIA Ampere architecture with Multi-Instance GPU (MIG) style partitioning and dedicated Tensor Cores for INT8 and FP16 inference, providing server-class AI throughput at edge power budgets. The 16 GB LPDDR5 memory delivers 102 GB/s bandwidth, sufficient to host multi-camera perception pipelines, transformer-based vision models, and generative AI workloads concurrently.
Typical applications include autonomous mobile robots (AMR), industrial machine vision, smart city traffic analytics, AI-powered video conferencing systems, and drone-based aerial inspection. The TAA/NDAA-compliant US country-of-origin designation makes this SKU particularly suited for US federal, defense, and government-procurement applications where non-TAA equivalents are ineligible.
When designing with this module, plan thermal dissipation around the configurable 10 W to 25 W TDP window and ensure the carrier board supplies the full PCIe Gen 4.0 reference-clock jitter budget. The 0A1 suffix denotes US COO; design teams should verify compliance requirements before substituting the 0A0 (Taiwan) or 000 (China) suffixes.
This page synthesizes distributor pricing, drop-in alternative module variants, and practical carrier-board design considerations not consolidated in the NVIDIA Jetson Orin NX datasheet.
直接替換替代品，適用於 900-13767-0000-0A1 — 相同封裝，腳位對腳位相容。需重新設計 PCB 的不同封裝零件不列入。
此系列的變體
與下列產品可直接替換的同系列型號： 900-13767-0000-0A1 （相同外形與佔位） — 差異在於 AI Performance, Camera Interfaces, Form Factor, GPU, Memory Bandwidth.
900-13767-0000-0A0
✅ 直接替換聯絡詢價
查看資料表 →900-13767-0000-000
✅ 直接替換📋 參考替代品（不在目錄中）
900-13767-0000-001
✅ 直接替換📋 參考替代品（不在目錄中）
900-13767-0000-0A1 規格（製造商公布）
| Module Type | System-on-Module (SOM) |
| Product Family | NVIDIA Jetson Orin NX |
| AI Performance | 100 TOPS (INT8) |
| GPU | 1024-core NVIDIA Ampere with Tensor Cores |
| CPU | 8-core Arm Cortex-A78AE 64-bit |
| Memory | 16 GB LPDDR5 |
| Memory Bandwidth | 102 GB/s |
| Storage | External via carrier (no on-module eMMC) |
| Video Decode | 2x NVDEC |
| Video Encode | 1x NVENC |
| PCIe | PCIe Gen 4.0 x4 |
| Camera Interfaces | 2x MIPI CSI-2 (4-lane) |
| Configurable TDP | 10 W to 25 W |
| Form Factor | 70 mm x 45 mm (260-pin SO-DIMM connector) |
| 工作溫度 | -25C to +80C (industrial modules vary) |
| TAA/NDAA Compliance | Yes (US Country of Origin) |
| COO Suffix | 0A1 = USA (0A0 = Taiwan, 000 = China per EMAC inc.) |
| RoHS 狀態 | Compliant (NVIDIA Jetson module product page) |
900-13767-0000-0A1 介面與連接器
900-13767-0000-0A1 依製造商公布提供以下介面與連接器：PCIe, 相機介面。擴充排針腳位、連接器腳位編號與電氣限制均定義於製造商文件中。
| PCIe | PCIe Gen 4.0 x4 |
| 相機介面 | 2x MIPI CSI-2 (4-lane) |
請參閱 製造商文件 以取得完整的擴充排針腳位定義與連接器腳位編號。
功率與熱特性
| 操作溫度 | -25C to +80C (industrial modules vary) |
900-13767-0000-0A1 由製造商公布的功率輸入與熱參數。
典型應用
900-13767-0000-0A1 適用於 6 應用s: Autonomous Mobile Robots (AMR), Federal/Defense Edge AI (TAA/NDAA), Industrial Machine Vision Systems, AI-Powered Video Conferencing, Smart City Traffic Analytics, Drone-Based Aerial Inspection.
Autonomous Mobile Robots (AMR)
The 900-13767-0000-0A1 Jetson Orin NX 16GB delivers 100 TOPS INT8 AI in a 70x45mm footprint that integrates cleanly into AMR mainboards where space and power are constrained. With 16 GB LPDDR5 and 102 GB/s bandwidth, the module simultaneously runs SLAM, multi-camera obstacle detection, and behavior-prediction transformer models without paging. The dual MIPI CSI-2 interfaces accept stereo vision front-end pairs typical of AMR payload bays. Configurable 10-25 W TDP supports battery-powered AGV designs with several hours of continuous operation.
推薦
Federal/Defense Edge AI (TAA/NDAA)
The 0A1 suffix designates US Country of Origin, satisfying Trade Agreement Act (TAA) and NDAA Section 889 sourcing requirements for federal procurement. This makes the module eligible for deployment in Department of Defense, DHS, and intelligence-agency programs where the non-TAA 900-13767-0000-000 variant would disqualify a bid. The 100 TOPS throughput supports real-time sensor fusion, RF anomaly detection, and on-device NLP workloads on forward-operating edge nodes. Per EMAC Inc. COO documentation, only the 0A1 variant meets Buy American Act qualification.
推薦
Industrial Machine Vision Systems
With 100 TOPS of INT8 throughput and dual 4-lane MIPI CSI-2 interfaces, the 900-13767-0000-0A1 hosts multi-camera defect detection pipelines at 30+ FPS on factory floors. The 16 GB LPDDR5 fits YOLOv8-large, EfficientDet, and transformer-based vision models without quantization throttling. The 70x45mm SOM form factor drops into standard PLC-side carrier boards from Auvidea and Connect Tech, accelerating factory automation retrofits. The module's TAA-compliance is essential for US aerospace and defense manufacturing inspection lines.
推薦
AI-Powered Video Conferencing
The 100 TOPS Jetson Orin NX 16GB module enables on-device AI camera features including auto-framing, speaker tracking, background replacement, and noise suppression without cloud round-trips. Its Ampere GPU runs NVIDIA Maxine SDK acceleration pipelines at sub-25W power envelopes typical of conference hardware. Dual MIPI CSI-2 interfaces support room-camera and presenter-camera streams simultaneously. The TAA-compliant 0A1 SKU makes it deployable in US federal conference rooms and DoD collaboration systems.
推薦
Smart City Traffic Analytics
Deploying the 900-13767-0000-0A1 in roadside intelligence cabinets enables multi-stream video analytics at intersections, including vehicle classification, pedestrian tracking, and incident detection on a single TAA-compliant edge node. The 100 TOPS throughput handles 4-8 simultaneous 1080p30 camera streams with detection inference, while the 16 GB LPDDR5 stores multi-model context buffers. NDAA/TAA compliance is mandatory for federally-funded smart-city deployments where non-compliant modules would disqualify grant-funded programs.
推薦
Drone-Based Aerial Inspection
The 70x45mm Jetson Orin NX 16GB module fits the gimbal-mount payload bays of commercial and defense UAVs where every gram matters. At 100 TOPS, it runs real-time object detection, thermal-anomaly mapping, and structural-crack analysis on-board without depending on operator downlink bandwidth. The configurable 10-25 W TDP lets mission planners tune AI throughput against battery endurance for inspection routes. The TAA-compliant 0A1 variant is mandatory for DoD and DHS drone programs requiring US-sourced compute modules.
推薦
推薦產品摘要
工程參考資料，適用於 900-13767-0000-0A1 — 比較、設計指引及合規資訊。
選型指南
與替代品比較
| 參數 | 本產品 | 900-13767-0000-0A0 | 900-13767-0000-000 | 900-13767-0000-001 |
|---|---|---|---|---|
| Brand | NVIDIA | NVIDIA | NVIDIA | NVIDIA |
| Package / Form Factor | Jetson Orin NX 70x45mm SOM (P3767, 260-pin SO-DIMM) | Jetson Orin NX 70x45mm SOM (P3767) - same | Jetson Orin NX 70x45mm SOM (P3767) - same | Jetson Orin NX 70x45mm SOM (P3767) - same |
| AI Performance (INT8) | 100 TOPS | 100 TOPS | 100 TOPS | 157 TOPS |
| GPU Cores | 1024-core Ampere | 1024-core Ampere | 1024-core Ampere | 1024-core Ampere |
| CPU | 8-core Arm Cortex-A78AE | 8-core Arm Cortex-A78AE | 8-core Arm Cortex-A78AE | 8-core Arm Cortex-A78AE |
| Memory | 16 GB LPDDR5 | 16 GB LPDDR5 | 16 GB LPDDR5 | 16 GB LPDDR5 |
| Country of Origin | USA (0A1 suffix) | Taiwan | China | China |
| TAA/NDAA Compliance | Yes (US COO) | No | No | No |
| Configurable TDP | 10 W to 25 W | 10 W to 25 W | 10 W to 25 W | 10 W to 25 W |
主要差異化特點
- US Country of Origin (TAA/NDAA compliant) (對比900-13767-0000-000)
- Higher AI throughput vs Taiwan COO alternate (對比900-13767-0000-001)
- Drop-in compatibility with existing Jetson Orin NX carriers (對比Qualcomm Robotics RB5 SOM)
設計說明
Estimated: at the 25 W upper TDP envelope in a sealed enclosure with 0.5 C/W heatsink and 5 C ambient, the junction temperature reaches approximately 25 + 25 * 4 = 125 C, exceeding the 105 C Jetson Orin NX thermal limit. Reference NVIDIA's Jetson Orin NX Thermal Design Guide: deploy a passive heatsink with at least 0.3 C/W theta-CA plus active airflow for sustained 25 W workloads, or throttle to 15 W mode in passively-cooled enclosures. The module exposes a PWM fan control output that should be tied to a thermistor on the heatsink.
The 260-pin SO-DIMM connector requires matched-impedance routing for PCIe Gen 4, USB 3.2 Gen 2, and MIPI CSI-2 lanes per NVIDIA's Jetson Orin NX Design Guide. Estimated: 85 ohm differential for PCIe/USB, 90 ohm differential for MIPI, with via count minimized to 2 per transition. Place the carrier-board voltage regulators within 25 mm of the SO-DIMM connector to minimize input-rail droop during 25 W load transients; use 22 uF + 1 uF + 0.1 uF decoupling at the module input pins.
Procurement teams must verify the COO suffix (0A1 vs 0A0 vs 000 vs 001) before accepting a delivery - the variants are visually identical and a non-TAA module delivered against a TAA-spec order will cause the entire shipment to be rejected by federal procurement. Per the NVIDIA Developer Forums thread on Jetson Orin NX part-number differences, NVIDIA support historically could not confirm cross-availability, so build COO verification into the receiving inspection process with barcode scans against the SAP purchase-order line item.
Configure nvpmodel to lock TDP at 15 W or 25 W before benchmarking - default 'auto' mode can hide thermal throttling that masquerades as algorithm performance variance. Reference the Jetson Orin NX power-estimation guide: a 25 W envelope draws up to 6.25 A at 4 V from the carrier's main 5 V or 12 V rail, so design the carrier's input protection (TVS, eFuse) for at least 8 A continuous. Use jetson_clocks to set GPU/CPU frequencies deterministically when comparing inference latency across modules.
MIPI CSI-2 lanes from the carrier's camera connectors to the Jetson SoM should use matched-length routing with skew below 50 ps across the 4-lane CSI group. PCIe Gen 4 reference-clock jitter budget is 100 fs RMS; use a clean 100 MHz clock generator on the carrier rather than feeding the SoM's recovered clock. Reference NVIDIA's Jetson Orin NX Pinmux Template and Design Guide for the validated reference carrier stack-up (8-layer, FR408HR, 1 oz copper outer / 0.5 oz inner).
合規資訊
RoHS/lead-free status per NVIDIA Jetson module product page. AEC-Q100 not applicable (SOM product category). TAA/NDAA compliance verified by EMAC Inc. COO documentation - 0A1 is US-built per the suffix.