Microsoft has built one of the top five publicly disclosed supercomputers in the world, making new infrastructure available in Azure for OpenAI to train extremely large artificial intelligence models, and taking a first step in that vein toward creating a platform for other organizations and developers to build upon.
Machine learning experts have historically built separate, smaller AI models that learn a single task, such as translating languages. But now AI is becoming a platform with general training that developers can start creating on, which will require supercomputing infrastructure – state-of-the-art hardware connected by high-bandwidth networks – and tools to train the models across those interconnected computers.
The supercomputer will enable “new applications that are hard to even imagine right now,” says Microsoft Chief Technical Officer Kevin Scott.