Sharp has begun accepting orders for AI servers for the Japanese market. It has established an implementation, operations and maintenance framework in collaboration with sales distributors and began accepting orders for NVIDIA RTX Pro servers in Japan today.
Sharp is entering the domestic AI server business to meet demand for AI infrastructure accompanying the spread of generative AI, and will begin accepting orders today, 15 September, with the aim of launching sales in fiscal 2026. The target customers include companies, local governments, research institutions and data center operators in Japan.
The products initially available are enterprise AI servers (NVIDIA RTX Pro servers). Sharp plans to begin accepting orders for AI servers for large-scale training and inference (NVIDIA HGX systems) in the future.
Sharp has reached basic agreements with 2 sales distributors, ダイワボウ情報システム and リョーサン菱洋, and will collaborate with トゥモロー・ネット, which provides operations and maintenance services, to establish a one-stop delivery framework. The companies will provide integrated support covering customer proposals, implementation, operations and maintenance in Japan.
AI servers are high-performance GPU servers that support inference, AI application development and the integration of AI into business systems. The enterprise model can accommodate up to 8 NVIDIA RTX Pro 6000 Blackwell Server Edition GPUs. The HGX system combines 8 GPUs using NVLink and adopts liquid cooling, with a design suited to the training and inference of large language models. Both systems use a high-performance network architecture based on NVIDIA's enterprise design philosophy.
Against a backdrop of growing demand for AI infrastructure development in Japan and global constraints on component supplies, Sharp will collaborate with 鴻海科技集団 (Foxconn) to strengthen its procurement and manufacturing capabilities. The specifications for the RTX Pro server are designated AI-B43100, allowing customers to select GPU, memory and storage configurations according to their intended applications. Orders for the HGX system are scheduled to open in fiscal 2027, and its specifications will be announced once finalized.
While demand for on-premises AI use is also increasing in Japan, the area in which services can be provided may be subject to geographic limitations.