Microsoft is expanding Azure’s AI and high-performance computing portfolio with new systems based on AMD’s latest processors and accelerators.
The announced lineup includes HDv2 virtual machines with almost 500 physical sixth-generation AMD EPYC cores, four terabytes of memory, local NVMe storage and high-bandwidth networking. Microsoft also outlined HXv2 systems for compute-intensive engineering workloads and an Azure configuration based on AMD’s Helios rack-scale architecture for AI inference.
The significance goes beyond a single cloud product launch. AI infrastructure buyers increasingly want alternatives across CPUs, accelerators and complete rack designs. That gives AMD an opening in a market where demand for training and inference capacity continues to outpace the ability of many organizations to deploy hardware themselves.
For Microsoft, a broader hardware portfolio can help Azure match different workloads with different price, memory and performance requirements. The competitive question will be how quickly these systems become available at scale and how easily customers can move existing software to the new configurations.
Source: Microsoft
