In the current wave of rapid development of AI, large language models have shown powerful intelligent interaction capabilities, but how to solve the many challenges they face when deployed on edge devices, such as computing power, memory, and energy consumption, and edge computing global implications has become the focus of industry attention. The NVIDIA Jetson series of intelligent edge computing technology products launched by Firefly provide reliable edge computing devices support and edge computing solutions for the deployment of large AI models on the edge with excellent edge computing capabilities and energy efficiency.
NVIDIA Jetson Orin Series
As the core module of these products, NVIDIA Jetson can provide accelerated computing capabilities suitable for various AI edge computing performance and price points. With its outstanding performance, excellent energy efficiency, and extremely easy development experience, it can meet the needs of various edge computing applications.
| Model | Jetson AGX Orin | Jetson Orin NX | Jetson Orin Nano |
| NPU | 275 TOPS | 157 TOPS | 67 TOPS |
| Power Consumption | 15W-60W | 10W-15W-25W-40W | 7W-15W-25W |
| Storage | 64GB eMMC 5.1 | External NVMe | External NVMe |
| Function | Flagship high-end edge computing platform, suitable for complex scenarios such as large-scale AI reasoning and multi-model parallel processing. | Lightweight, high-performance edge devices that balance computing power and size/power consumption requirements. | Entry-level edge AI development kit, focusing on low cost and ease of use |
| Application | Autonomous driving, industrial robots, high-performance edge servers, edge computing solutions, and other fields that require extremely high computing power and concurrency. | Medium-sized AI applications that are sensitive to size and power efficiency, such as smart cameras, drones, and portable medical devices. | Low-complexity AI scenarios such as educational experiments, prototype verification, and IoT devices. |
NVIDIA Jetson Embedded Systems
Enabling edge compute use cases and robotics application development with a comprehensive AI software stack and ecosystem powered by generative AI at the edge and the NVIDIA Metropolis and Isaac™ platforms.
- Rapidly build AI end-to-end solutions.
- Software stack and ecosystem to facilitate easy development.
- Support cloud-native technologies and workflows such as containerization and orchestration.
NVIDIA Jetson and DeepSeek
| Deepseek Model | Jetson Core Module |
| DeepSeek R1 Qwen-1.5B | AGX Orin 64GB, Orin NX16GB, Orin Nano 8GB |
| DeepSeek R1 Qwen-7B | AGX Orin 64GB, Orin NX16GB, Orin Nano 8GB |
| DeepSeek R1 Llama-8B | AGX Orin 64GB, Orin NX16GB, Orin Nano 8GB |
| DeepSeek R1 Qwen-32B | AGX Orin 64GB |
| DeepSeek R1 Llama-70B | AGX Orin 64GB |
Among many AI models, DeepSeek stands out for its excellent language understanding and generation capabilities and demonstrates strong reasoning capabilities in multimodal scenarios. By combining NVIDIA Jetson's high computing power and low latency, DeepSeek can complete reasoning tasks on the device side, which can not only reduce response time but also enhance privacy protection through localized data processing, helping to upgrade the intelligence of edge devices in more industries.
Firefly NVIDIA Jetson Series
AI Edge Computing Server
- Up to 10 computing nodes can be configured, each equipped with an NVIDIA Jetson Orin series processor, with a total computing power of up to 1570 TOPS, providing strong computing power support for artificial intelligence and deep learning.
AI Embedded Computer Hardware
- "Small box" embedded computer devices that can hold large AI models can realize comprehensive private deployment of artificial intelligence and adapt to various AI application scenarios.
NVIDIA Jetson Orin Development Kit
- NVIDIA Jetson AGX Orin/Jetson Orin Nano/Jetson Orin NX are compact, powerful supercomputers that deliver strong AI performance and can seamlessly run generative AI models.
NVIDIA Jetson Orin Computing Module
- NVIDIA Jetson Orin modules have computing capabilities of different performance levels, and their modular design supports hardware compatibility and rapid deployment to meet the needs of a variety of autonomous applications.
Edge Computing Examples
Edge computing architecture is widely used in smart cities, healthcare, retail, factory logistics, industry, construction, and other industries.
