What Is the BOXER-8651AI?
The BOXER-8651AI is a compact, fanless embedded AI system built by AAEON around the NVIDIA® Jetson Orin™ NX module. It is designed for edge AI deployments where a small, rugged, ventless enclosure must deliver accelerated deep learning inference close to the data source rather than in a distant data centre. A newer variant, the BOXER-8651AI-PLUS, has also been introduced with enhanced AI performance and an integrated software security framework.
Key Specifications
The system pairs an NVIDIA Ampere architecture GPU — with 1024 CUDA® cores and 32 Tensor Cores — with the Jetson Orin NX module, which includes a dedicated Deep Learning Accelerator. The base BOXER-8651AI is rated for up to 70 TOPS of AI performance, while the PLUS variant, available with 8GB or 16GB Jetson Orin NX modules in Super Mode, is rated up to 157 TOPS with NVIDIA JetPack™ 6.2 support. The base unit supports JetPack 5.0 and above. Measuring roughly 105mm x 90mm x 52mm, it is among the smallest fanless systems engineered for the Orin NX.
| Attribute | BOXER-8651AI | BOXER-8651AI-PLUS |
|---|---|---|
| AI Module | NVIDIA Jetson Orin NX | NVIDIA Jetson Orin NX (8GB/16GB, Super Mode) |
| AI Performance | Up to 70 TOPS | Up to 157 TOPS |
| GPU | Ampere, 1024 CUDA + 32 Tensor Cores | Ampere, 1024 CUDA + 32 Tensor Cores |
| JetPack Support | JetPack 5.0 and above | JetPack 6.2 |
| Cooling | Fanless | Fanless |
| Footprint | ~105 x 90 x 52 mm | Compact embedded form |
Use Cases and Applications
Edge AI systems of this class are typically used for real-time video analytics, machine vision on production lines, smart camera and surveillance inference, autonomous mobile robots, and AI-assisted inspection in industrial or outdoor environments. The fanless, ventless chassis suits dusty, hot, or vibration-prone sites where a fan would be a maintenance liability. Developers build on the NVIDIA Jetson software stack, so CUDA, TensorRT, and JetPack tooling carry over from prototype to deployment.
Choosing an Edge AI Platform
When specifying an edge AI computer, consider the TOPS budget your models actually need, the memory footprint of your neural networks, the operating temperature range, mounting and I/O requirements, and long-term software support. Also weigh the platform ecosystem: NVIDIA Jetson offers mature acceleration libraries, while x86 industrial PCs offer familiarity with standard Windows and Linux application stacks and broad peripheral compatibility.
Thinvent's Products for Edge and Industrial Computing
Thinvent manufactures industrial computers, mini PCs, thin clients, and all-in-one PCs for deployment worldwide. Our fanless industrial PC range — including systems built on Intel® Core™ i3-1215U processors with 6 cores, up to 4.4 GHz, 10 MB cache, and 16GB DDR4 memory — is suited to edge data collection, kiosk control, gateway duties, HMI, and shop-floor workloads that pair well with dedicated AI accelerators. For AI-heavy inference, Thinvent platforms can act as the host, gateway, or management layer alongside Jetson-class edge devices, giving integrators a single, consistent hardware supply for the whole installation.
