The day before yesterday, Sharp unveiled its strategic partnership with Foxconn, aiming to make significant inroads into the AI server market. The collaboration has set its sights on introducing two innovative products by 2027, specifically targeting a broad spectrum of customers including enterprises, the Japanese government, research institutions, and data centers.

The inaugural product is a modular server tailored for the enterprise AI market. It boasts the capability to accommodate up to 8 NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, offering a high degree of customization. Under the hood, it features dual Intel Xeon 6 series processors, a robust 3TB memory capacity, a substantial 38.4TB of hard drive storage, and is powered by 4 x 3200W power supplies, ensuring optimal performance and reliability.

The second offering is geared towards the training and inference of large AI models. It is outfitted with the NVIDIA HGX Rubin NVL8, which seamlessly integrates 8 GPUs through NVLink technology. Additionally, it employs liquid cooling technology to maintain optimal operating temperatures and supports NVIDIA ConnectX-9 SuperNIC and NVIDIA BlueField-4 DPU configurations, enhancing its overall efficiency and versatility in handling complex AI workloads.