Sugon Teases World's First 64-Thread Mobile Workstation for Local AI Inference
Sugon has teased the world's first 64-thread mobile workstation, featuring an ultra-slim 16.9mm chassis and 16GB of VRAM. The device is designed to run local 35-billion parameter Mixture of Experts (MoE) AI models at speeds of up to 50 tokens per second. Packing 64 threads into a slim laptop form factor allows developers and engineers to execute large AI models locally without relying on cloud infrastructure. If performance claims hold true, its 3x to 5x speed advantage over mainstream hardware could accelerate portable edge AI computing. Sugon claims the system provides 3 to 5 times the AI inference speed of current mainstream industry hardware while keeping the chassis under 17mm thick. However, specific details about the underlying processor architecture, GPU chipset, and thermal cooling solution have not yet been disclosed.
## BACKGROUND
Sugon is a major Chinese high-performance computing manufacturer specializing in supercomputers and enterprise servers. Mixture of Experts (MoE) is a machine learning architecture that divides a model into multiple specialized sub-networks, enabling large-scale models to process data faster and with lower computational cost.