General Computing
The FusionPoD rack-scale liquid-cooled server supports a centralized power supply ranging from 33 kW to 105 kW with 64 to 144 CPUs per cabinet. The node adopts a three-bus (liquid, network, and power) blind mating design, enabling cable-free maintenance. It can be widely deployed in telecom, Internet company, government and enterprise data centers, as well as Internet Data Centers (IDCs).
AI Computing
The FusionPoD for AI rack-scale liquid-cooled server supports a centralized power supply ranging from 105 kW to 240 kW with 64 to 128 accelerator cards per cabinet. It is compatible with over 10 types of computing capabilities, making it ideal for large‑scale and ultra‑large‑scale model training, inference, and other AI scenarios.
The FusionPoD rack-scale liquid-cooled server supports a centralized power supply ranging from 33 kW to 105 kW with 64 to 144 CPUs per cabinet. The node adopts a three-bus (liquid, network, and power) blind mating design, enabling cable-free maintenance. It can be widely deployed in telecom, Internet company, government and enterprise data centers, as well as Internet Data Centers (IDCs).
The FusionPoD for AI rack-scale liquid-cooled server supports a centralized power supply ranging from 105 kW to 240 kW with 64 to 128 accelerator cards per cabinet. It is compatible with over 10 types of computing capabilities, making it ideal for large‑scale and ultra‑large‑scale model training, inference, and other AI scenarios.
