Super X AI Technology Limited has unveiled its latest flagship product, the SuperX XN9160-B200 AI Server. This next-generation system is engineered to address the escalating demand for scalable, high-performance computing power. The server integrates NVIDIA’s Blackwell B200 GPU architecture, positioning it for intensive workloads in AI training, machine learning, and high-performance computing (HPC).
The XN9160-B200 AI Server is purpose-built to accelerate large-scale distributed AI training and inference tasks. The system’s performance capabilities are presented as a significant advancement, reportedly offering up to a 30-fold acceleration in AI innovation when compared to the previous H100 series. Its computational output is designed to rival that of a traditional supercomputer while being delivered in a more compact, enterprise-grade form factor. This launch represents a key development in the company’s AI infrastructure strategy.
At the core of the server are eight NVIDIA Blackwell B200 GPUs connected via fifth-generation NVLink technology, which facilitates an inter-GPU bandwidth of up to 1.8TB/s. This configuration is designed to achieve up to a threefold improvement in the speed of large-scale AI model training. For inference tasks, the system leverages 1440 GB of high-bandwidth HBM3E memory, demonstrating a substantial increase in performance over the prior generation. The system’s computational architecture is further supported by 6th Gen Intel® Xeon® processors and high-speed DDR5 memory, housed within a 10U chassis.
The server is optimized for a range of advanced applications. These include the training and inference of foundation models using techniques such as reinforcement learning and distillation, multimodal model processing, and complex HPC simulations. Specific fields of application include climate modeling, drug discovery, seismic analysis, and insurance risk modeling.
To ensure operational stability for critical AI missions, the XN9160-B200 incorporates an advanced multi-path power redundancy solution, featuring multiple redundant power supplies to mitigate the risk of single-point failures. Management is handled by a built-in intelligent management system that allows for remote monitoring. The company notes that each server undergoes extensive full-load stress testing and aging screening to ensure reliability.
The primary markets for the XN9160-B200 include large technology companies, academic and research institutions, and specialized sectors such as finance, pharmaceuticals, and government meteorological agencies. Super X AI Technology Limited, headquartered in Singapore, provides a portfolio of hardware, software, and end-to-end services for AI data centers to a global institutional clientele.
