Inventec is launching its P8000IG6 server for rapidly and flexibly scaling up data center capacity to take on the most advanced AI and high-performance computing (HPC) workloads. As new AI technologies proliferate, demand for data centers is growing quickly. According to McKinsey, data center operators are caught between the need for more capacity and rising capital costs, resulting in more data centers relying on prefabricated and modular solutions to expand while reducing setup time.

Powered by up to eight GPUs, P8000IG6 is built around the NVIDIA® HGX? AI supercomputing platform, optimized for the most advanced AI and HPC workloads, and customers can select from NVIDIA® H100, H200, or B100 GPUs. To ensure seamless interconnectivity, NVIDIA® NVLink® and NVSwitch® accelerate communication among all eight onboard GPUs, while Inventec's switchboard bridges the server's GPUs with others in the data center and offers maximum bandwidth for GPU Direct RDMA.

With its modular design, the server enables rapid deployment and installation, facilitates expansion and upgrading existing infrastructure, and provides extensive compatibility for use with other types of servers. Inventec's experts also offer white-glove customization for data centers in need of specialized solutions. Adding to its array of top technologies, P8000IG6 offers the latest DDR5 SDRAM and PCIe Gen.

5.0 connectivity. To boost power efficiency and reliability, its CPU and GPU modules have decoupled power sources ? 12V and 54V, respectively.

In addition, both air and liquid cooling options further expand its customizability. On top of rapid deployment and seamless expansion, the modular design of the P8000IG6 streamlines maintenance. The CPU and GPU are separated into different trays, and there are several modules ?

including for the fans, power supply, and NIC ? combining with the tool-less design for service and maintenance made simple.