275 TOPS in 16.8mmThe Impossible Made Possible
Powered by NVIDIA Jetson AGX Orin 64GB delivering 275 TOPS with 2048-core Ampere GPU, 64GB unified memory, and 1PB storage. Ultra-thin design at just 485g.
Revolutionary Design
XaNode AI Micro Server
Actual size: 12.0" x 3.0" x 0.66" • Weight: 485g
Detailed Specifications
Breakthrough performance metrics that redefine edge AI computing
275
TOPS per node
Processing Power
NVIDIA Jetson AGX Orin 64GB with 2048-core Ampere architecture GPU, 64 Tensor Cores, max GPU frequency 1.3GHz, and 12-core ARM Cortex-A78AE CPU.
16.8
mm thick
Form Factor
12.0" x 3.0" x 0.66" - Ultra-thin design at just 485g. Revolutionary engineering in compact form factor.
64
GB LPDDR5
Unified Memory
64GB 256-bit LPDDR5 unified memory architecture with 204.8GB/s bandwidth for seamless CPU-GPU data sharing.
1
PB Storage
Storage Capacity
Massive 1PB storage capacity per node with NVMe performance for datasets and model storage.
2×100
Gb/s NICs
Network Connectivity
Dual 100Gb/s Network Interface Cards (NICs) provide high-speed connectivity for distributed AI workloads and data transfer.
<200
ns latency
CXL 3.0 Fabric
First edge AI server with CXL 3.0 enabling coherent memory sharing and pooling across nodes.
Performance Benchmarks
Industry-leading AI inference performance that redefines edge computing capabilities
01
ResNet-50 Performance
Achieve 14,500 fps for image classification tasks, enabling real-time video processing and analysis at unprecedented speeds for edge deployments.
02
YOLOv5s Object Detection
Process 2,800 fps for object detection, perfect for autonomous vehicles, security systems, and industrial inspection applications.
03
BERT-Base NLP
Generate 8,200 tokens per second for natural language processing, enabling real-time conversational AI and text analysis at the edge.
04
Density Advantage
42 XaNodes in a single 2U XaPod deliver 11,000 TOPS total - 30x more performance per rack unit than traditional GPU servers.
01
ResNet-50 Performance
Achieve 14,500 fps for image classification tasks, enabling real-time video processing and analysis at unprecedented speeds for edge deployments.
02
YOLOv5s Object Detection
Process 2,800 fps for object detection, perfect for autonomous vehicles, security systems, and industrial inspection applications.
03
BERT-Base NLP
Generate 8,200 tokens per second for natural language processing, enabling real-time conversational AI and text analysis at the edge.
04
Density Advantage
42 XaNodes in a single 2U XaPod deliver 11,000 TOPS total - 30x more performance per rack unit than traditional GPU servers.
Experience XaNode Performance
Join the edge AI revolution with the world's most powerful ultra-thin AI server.