SK Biopharmaceuticals deploys NVIDIA B200 AI cluster as Sugon announces 64-thread workstation

SK Biopharmaceuticals has announced the deployment of 128 NVIDIA Blackwell B200 AI accelerators to build a large-scale proprietary AI computing infrastructure. The system reaches a total computing power of approximately 0.58 exaflops at FP8 precision. The company intends to use this infrastructure to develop foundational models for biomedicine and optimize business processes, specifically targeting the “translational research AI transformation” project to accelerate drug discovery and target research.

In a separate development, Chinese manufacturer Sugon is reportedly preparing to launch a high-performance, ultra-thin mobile workstation. The device is expected to be the world’s first mobile workstation featuring 64 threads, potentially utilizing Hygon C86 processors. Designed for local AI tasks, the workstation is claimed to handle a 35-billion parameter Mixture of Experts (MoE) model at an inference speed of approximately 50 tokens per second. This hardware aims to provide professional-grade AI development and large-scale computing capabilities in a portable form factor.