During the build conference, Microsoft announced it’s team up with OpenAI to create one of the fastest supercomputers on the Azure framework. Microsoft invested $1 billion dollars in this partnership. The 285,000-core machine, Azure, would have ranked in the top five of the supercomputer rankings.
Microsoft doesn’t tell any more than this, except that the specs are going to have 10,000 GPU’s and 400 gigabits per second for their network connectivity on each server. To be counted on one of the top five supercomputers, a machine will need to reach speeds of over 23,000 teraflops each second.
Currently, the number one spot is helped by IBM, and they have over 148,000 teraflops per second for their machines. This leaves quite a large margin to make the top five between IBM and the server in the fifth spot.
Since the investment, OpenAI chose to make Azure its cloud of choice for the supercomputer they are developing. OpenAI made a name for itself by training for very large models.
CEO Sam Altman stated, “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 out dream system, what would it look like?'” After this, him and his team decided to say that Microsoft was able to build it.
Do you want to publish on Apple News, Google News, and more? Join our writing community, improve your writing skills, and be read by hundreds of thousands around the world!