ASUS has announced that its high-density AI server, the ESC A8A-E12U, now supports the newly launched AMD Instinct MI350 series GPUs. This update aims to provide enterprises, research institutions, and cloud providers with enhanced performance and a seamless upgrade path for next-generation artificial intelligence (AI) and high-performance computing (HPC) workloads.
The addition of AMD’s latest GPUs enables faster, more energy-efficient execution of large AI models and scientific simulations. It also ensures compatibility with existing systems, making it easier for organisations to upgrade their infrastructure without significant overhaul.
Enhanced performance and simplified scalability
Built on the 4th Gen AMD CDNA architecture, the Instinct MI350 series GPUs deliver 288GB of HBM3E memory and bandwidth of up to 8TB/s per GPU. This allows the ESC A8A-E12U to tackle more complex workloads while using fewer GPUs, lowering server density and simplifying deployment planning.
The MI350 series also supports low-precision compute formats such as FP4 and FP6, which accelerate performance in generative AI, inference, and machine learning applications. Importantly, these GPUs are drop-in compatible with systems based on previous models, such as the MI325X, offering a cost-effective and efficient upgrade path.
ASUS’s ESC A8A-E12U is a 7U server that pairs the MI350 series GPUs with dual AMD EPYC 9005 processors. This configuration supports intensive tasks such as training large language models, fine-tuning generative AI models, and conducting scientific research simulations. The improved memory bandwidth and capacity mean fewer GPUs are required to run larger models, which translates into reduced energy use and lower total cost of ownership for data centres.
Built-in security for multi-tenant environments
Security is another area where the new configuration stands out. The MI350 series GPUs include features such as Secure Boot, Device Identifier Composition Engine (DICE) attestation, single-root I/O virtualisation (SR-IOV), and encrypted GPU-to-GPU communication. These capabilities support secure multi-tenant operations, making the system suitable for environments that demand high levels of data protection, such as finance, government, and healthcare.
Paul Ju, General Manager of the ASUS Server Business Unit, stated: “As AI workloads continue to evolve in complexity and scale, ASUS is committed to delivering infrastructure solutions that help our customers move faster. With support for AMD Instinct MI350 series GPUs, our ESC A8A-E12U server provides a robust, future-ready platform for next-era AI and HPC deployments.”
Continued focus on enterprise innovation
ASUS’s move to integrate the latest GPU technology strengthens its position as a key player in enterprise and cloud infrastructure. The ESC A8A-E12U reflects the company’s focus on building scalable and secure platforms capable of handling the demands of modern data centres.
By offering enhanced performance, energy efficiency, and built-in security, the updated ESC A8A-E12U positions ASUS to meet the needs of organisations pushing the boundaries in AI and HPC.
