According to Meta representatives, Azure will be used to accelerate research and development across the Meta AI group. Meta also plans to use a dedicated Azure cluster of 5,400 GPUs that use Azure’s latest virtual machine (VM) series in Azure for some of their extensive AI research.
Microsoft and Meta also announced that they will work together to scale PyTouch adoption on Azure.
The AI partnership between Meta and Microsoft began in 2017 when they announced the ONXX (Open Neural Network Exchange) format in 2017.
The project aims to enable developers to move deep-learning models between different AI frameworks. In 2018, Microsoft open sourced the ONXX Runtime, which is the inference engine for models in the ONXX format.
In 2021, Meta started using Azure VMs for some of its large-scale AI research, particularly distributed AI training. Also in 2021, Microsoft publicly committed to increasing its support for enterprise customers using Meta’s PyTorch deep-learning framework on Azure.