Advantech Edge AI Systems Built on NVIDIA RTX PRO Blackwell GPUs Empower Enterprise On-Premises LLM Applications

Advantech, Taipei, Taiwan 13.06.2025 - Generative AI technology is rapidly advancing toward edge applications, driving significant growth in enterprise AI computing demands. As Large Language Models (LLMs) evolve, more businesses are choosing to train LLMs internally, not only to enhance information protection but also to flexibly adapt models according to business needs for optimal results. However, how to increase computing power while reducing energy consumption and total cost of ownership (TCO) has become a core concern for enterprises. 

Advantech's IPC-730 integrated with the NVIDIA RTX PRO 6000 Blackwell Workstation Edition GPUs provides enterprises with a high-performance platform for LLM training and inference. The new NVIDIA RTX PRO Blackwell GPUs deliver twice the memory capacity compared to the previous generation along with enhanced CUDA computing performance, ensuring AI calculations run Advantech's edge AI solution more efficiently with lower energy consumption per token.

IPC-730 Enhances AI Computing Performance with integrated 600W NVIDIA RTX PRO 6000 Blackwell GPU 

NVIDIA Blackwell architecture breaks through traditional computing bottlenecks, delivering superior performance with lower energy consumption. Advantech's edge computing systems include high-performance industrial computers IPC-730 and servers that fully integrate the latest NVIDIA latest Blackwell series GPUs, including NVIDIA RTX PRO 6000 Blackwell Server Edition, NVIDIA RTX PRO 6000 Blackwell Workstation Edition, and NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition, enabling enterprises to flexibly deploy AI applications and create high-performance, low-power AI computing solutions.

The NVIDIA RTX PRO Blackwell series GPUs feature a second-generation Transformer Engine that significantly improves AI training and inference efficiency. With a dual-slot design supporting up to 96GB of memory per GPU, they ensure stability and performance for AI computing. Compared to the previous generation GPUs, memory capacity has doubled, enabling support for more complex language models while reducing power consumption per token by 50% at the same computational performance, significantly lowering TCO. The industrial-grade durable design is suitable for long-term demanding computational tasks, ensuring stable system operation. Advantech can also provide IPC or server solutions based on different application requirements, meeting diverse enterprise AI computing needs.

Advantech's edge AI solution satisfies various AI application scenarios. Enterprise AI information analysis supports real-time data analysis and decision-making, improving operational efficiency; smart healthcare applications cover medical record analysis, image diagnosis, and clinical decision support, further enhancing medical AI accuracy; automated customer service systems optimize interactive experiences through LLM technology, improving response speed and reducing operational costs.

Advantech Edge AI Solution Test Data Validates Excellent Performance

Advantech's edge AI solutions have demonstrated excellent performance in multiple tests. In LLM computing scenarios, token processing efficiency is a key factor affecting AI model performance, including Eval Rate, Prompt Eval Rate, and VRAM usage. These metrics determine the real-time response and operational efficiency of the model. Through the powerful computing capabilities of NVIDIA RTX PRO Blackwell GPUs, enterprises can accelerate token processing, optimize inference and training performance of language models, ensure optimal AI performance in language understanding and content generation applications, and further enhance the competitiveness of edge AI solutions.

System model: IPC-730+AIMB-789+ RTX PRO 6000 Blackwell Workstation edition / RTX 6000Ada

In 3DMark Time Spy Graphics benchmark testing, the RTX PRO 6000 Blackwell Workstation Edition outperforms the previous-generation RTX 6000 Ada by over 53%. This substantial leap in rendering power enables professionals to tackle complex visualization, real-time ray tracing, and high-fidelity content creation with greater speed and efficiency.

The RTX PRO 6000 Blackwell Workstation Edition, built on NVIDIA’s latest Blackwell architecture, delivers a transformative leap in generative AI performance. In both image and text generation tasks, the RTX PRO 6000 Blackwell Workstation Edition outpaces the previous Ada-based model with up to 77% faster image generation speed and up to 37% higher token output throughput in large language models. Latency is significantly reduced across key AI workloads. For instance, Stable Diffusion FP16 image generation time is cut by over 54%, while UNET processing speed increases by 47%, enabling smoother and more responsive AI experiences across edge and cloud environments.

Advantech Accelerates Enterprise Adoption of Next-Generation LLM Computing

As an NVIDIA Elite Partner, Advantech continues to advance Edge AI computing technology by developing industrial-grade high-performance computers and servers integrated with GPUs, helping enterprises deploy LLM applications and providing more flexible, energy-efficient, and cost-effective edge AI computing solutions. All products are now available for order.