Inventec P9000AG7 HGX B200 NVL8 180GB Air-Cooled (8) AMD Turin (2) AI Graphics Card GPU Server

The Inventec P9000AG7 (AC) is purpose-built to meet the rising demands of advanced AI and HPC workloads that require exceptional scalability, compute density, and energy efficiency. By integrating the latest AMD EPYC™ 9005 Series processors with the NVIDIA HGX™ B200 8-GPU system, it delivers a balanced architecture that combines powerful processing with seamless scalability across diverse deployment needs. This design enables enterprises to handle large-scale datasets efficiently, support complex model development, and drive accelerated computing with confidence. It offers a resilient and high-performance platform ready to tackle the challenges of AI training, scientific computing, and data-intensive operations.

Flexible Compute Engine for Evolving AI Workloads

The P9000AG7 (AC) features a powerful and flexible 10U air-cooled design, integrating dual AMD EPYC™ 9005 Series processors with the NVIDIA HGX™ B200 8-GPU platform, delivering the compute performance required for demanding AI training and inference workloads. It supports up to 12 U.2 NVMe SSDs, which are connected under the same switchboard as the GPUs to keep large datasets close to compute resources and accelerate data movement during intensive workload processing, enabling optimal NVIDIA GPUDirect® Storage (GDS) performance. An advanced air-cooling system maintains stable thermal performance even during sustained high-load GPU operations, empowering organizations to scale data-driven workloads with confidence.

Powering Seamless Data Movement for Next-Gen AI

The P9000AG7 (AC) keeps pace with the needs of evolving AI models by supporting high-speed DDR5 memory, delivering up to 6400 MT/s at 1DPC to ensure fast data access for compute-intensive workloads. The NVIDIA NVLink™ Bridge connects GPUs for seamless inter-GPU communication, supporting parallel computing across demanding tasks. Additionally, the Inventec-designed switchboard is optimized to provide maximum bandwidth for NVIDIA GPUDirect® RDMA, enabling direct, high-speed data transfers between GPUs and other system components, which reduces latency and accelerates data movement throughout the system. This architecture ensures that the P9000AG7 (AC) can efficiently handle complex AI training, large-scale simulations, and data-heavy HPC applications with consistent performance.

Modular Design for Simplified Serviceability

Maintaining peak system performance requires thoughtful design. The P9000AG7 (AC) adopts a modular approach that simplifies serviceability with tool-less access to fans, power supplies, NIC, and storages. The clear separation between the CPU and GPU modules improves airflow management and thermal zoning, while also making upgrades and maintenance straightforward to reduce downtime in mission-critical AI environments.

P9000AG7 (AC)