Compal unveils 3 new NVIDIA MGX servers for enterprise AI and HPC workloads

Compal’s new NVIDIA MGX architecture-based server platforms are reshaping the enterprise AI and HPC computing landscape with unprecedented computational power and flexibility. Unveiled at GTC 2025, these three new server models represent a significant advancement in data center technology, offering tailored configurations for various high-performance computing needs while leveraging NVIDIA’s latest GPU innovations to address the growing demands of AI workloads and scientific computing applications.

The big picture: Compal Electronics has launched three new server platforms built on NVIDIA MGX architecture, designed specifically for enterprise-level AI, HPC, and high-load computing applications.

• The lineup includes the SX420-2A 4U AI Server, SX224-2A 2U AI Server, and SX220-1N 2U AI Server, each targeting different segments of the high-performance computing market.
• These servers incorporate NVIDIA’s latest GPU technology, including support for RTX PRO 6000 Blackwell GPUs and the NVIDIA GH200 Grace Hopper Superchip.

Key details: The flagship SX420-2A 4U AI Server features flexible rack design supporting both industry standard 19″ and ORV3 21″ configurations.

• It can be configured with up to 8 RTX PRO 6000 Blackwell GPUs, significantly enhancing data center compute performance and resource utilization.
• The server is specifically engineered for deep AI-HPC applications, providing the computational foundation for advanced artificial intelligence workloads.

Technical innovations: The SX224-2A 2U AI Server integrates NVIDIA MGX architecture with AMD x86 platform technology for versatile configuration options.

• The design is future-compatible and optimized for diverse computing workloads spanning AI-HPC and AI-Graphics applications.
• This platform enables tailored performance adjustments to meet specific computational requirements.

Advanced GPU capabilities: The RTX PRO 6000 Blackwell GPU included in these servers delivers exceptional performance with 96GB of GDDR7 memory.

• The passively cooled thermal design ensures stable operation even under extreme computational loads.
• These GPUs provide acceleration for both agentic and physical AI applications, as well as scientific computing, graphics, and video workloads.

Specialized computing solutions: The SX220-1N 2U AI Server is engineered specifically for giant-scale AI and HPC applications requiring massive computational resources.

• It features the NVIDIA GH200 Grace Hopper Superchip and employs NVIDIA NVLink-C2C technology to deliver a coherent memory pool.
• This architecture enables faster memory speeds and exceptional bandwidth to handle large-scale computational tasks.

Why this matters: The comprehensive server lineup represents a significant advancement in enterprise computing technology, addressing the growing demands for specialized AI and HPC infrastructure.

• These platforms satisfy the rigorous requirements of diverse data center applications while offering the flexibility to adapt to evolving computational needs.
• As AI workloads continue to grow in complexity and scale, purpose-built hardware solutions like these become increasingly critical for organizational performance and efficiency.