Microsoft has built a new supercomputer exclusively for OpenAI, designed to train large-scale artificial intelligence (AI) models, the company confirmed at its Build developers conference.
At the event, Microsoft said the 285,000-core device, built on top of Azure infrastructure, places among the top five fastest supercomputers in world.
According to TechCrunch, the device - which also boasts 10,000 GPUs and 400 gigabits per second of network connectivity per server - would need to achieve at least 23,000 teraflops per second to rank as the fifth fastest device.
“The computer is connected to Azure, but is a dedicated resource of OpenAI’s. They paid for the system, paying both Microsoft and other suppliers. The total cost is not being disclosed,” a Microsoft spokesperson told TechCrunch.
Although the supercomputer itself was built exclusively for OpenAI, Microsoft plans to make its own large AI models and training optimisation tools available via Azure AI services and GitHub.
For clients looking to take advantage of AI facilities, but that do not need a dedicated supercomputer, the Azure AI offering also includes AI accelerators found in the new supercomputer.
“As we’ve learned more and more about what we need and the different limits of all the components that make up a supercomputer, we were really able to say, ‘If we could design our dream system, what would it look like?’” said OpenAI CEO Sam Altman. “And then Microsoft was able to build it.”