Sharp announced two days ago its entry into the AI server market, with new products planned to be sold starting in 2027, targeting customers such as enterprises, the Japanese government, research institutions, and data centers. For this launch, Sharp has partnered with Foxconn to introduce two AI server products.

Enterprise Model: Modular Design, Up to 8 GPUs

The first product is aimed at the enterprise AI market, featuring a modular design that can accommodate up to 8 NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. Customers can customize various configurations. In terms of specifications, this enterprise server uses dual Intel Xeon 6 series processors, equipped with DDR5 RDIMM 96GB × 32 (totaling 3TB) memory and 38.4TB hard disk space, along with four 3200W power supplies.

Training Model: HGX Rubin NVL8 with Liquid Cooling

At the same time, Sharp will also launch an AI server designed for large AI model training and inference. This product is equipped with NVIDIA HGX Rubin NVL8, using NVLink to integrate 8 GPUs into one server. It features liquid cooling and supports configurations with NVIDIA ConnectX-9 SuperNIC and NVIDIA BlueField-4 DPU.