Hardware & InferenceBusiness & Market 🇺🇸 24.07.2026 07:04

Microsoft Expands Azure Infrastructure for AI and HPC with AMD

MicrosoftMicrosoft
Microsoft has announced three new series of Azure virtual machines based on the latest AMD EPYC processors and the AMD Helios platform: HDv2 for data processing, HXv2 for chip design and technical computing, and ND MI455X v7 for AI inference. These solutions aim to deliver performance and efficiency for growing AI workloads, including large agentic systems and high-performance computing.
Microsoft announced the expansion of its Azure infrastructure for artificial intelligence (AI) and high-performance computing (HPC) using AMD solutions. The company is integrating the AMD Helios platform and next-generation AMD EPYC processors into three new types of virtual machines (VMs): HDv2 for data processing, HXv2 for electronic design automation (EDA) and other technical computing, and ND MI455X v7 for AI inference. HDv2 VMs, developed jointly with AMD, contain nearly 500 cores of 6th generation AMD EPYC, 4 TB of RAM, 32 TB of local NVMe storage, and a 400 Gbps Azure Boost network, enabling efficient data preparation, search, reinforcement learning, and agent coordination. HXv2 VMs are equipped with 176 AMD EPYC cores clocked at over 5 GHz, 3D V-cache technology, up to 4 TB of RAM, and an 800 Gbps InfiniBand network, accelerating RTL simulations, scientific and engineering calculations. Microsoft is also collaborating with Synopsys to optimize AI EDA solutions on Azure. The ND MI455X v7, based on AMD Helios, is designed for large-scale inference, including reasoning, search, and agent workloads. Thus, Microsoft is expanding the choice of compute resources for customers, allowing them to select the optimal configuration for specific AI tasks.
Source: Microsoft AI — original
Our earlier posts on this topic ↓
Fresh news