Microsoft designed a new supercomputer for artificial intelligence out of tens of thousands of chips

No time to read?
Get a summary

American company Microsoft has built a state-of-the-art supercomputer for its AI research startup OpenAI. Agency reports Bloomberg.

It is stated that Microsoft did not have anything similar to what was required for the development of OpenAI at the time of signing a contract with a startup.

“Microsoft had to find ways to combine tens of thousands of Nvidia’s A100 graphics chips to train AI models, as well as change how servers were placed in racks to avoid power outages,” the article states.

Microsoft Vice President Scott Guthrie, which oversees cloud computing and artificial intelligence, did not specify the specific cost of the project, only that the amount would likely exceed several hundred million dollars.

“We have built a system architecture that can operate at a massive scale and be reliable. This is what made ChatGPT possible,” explained Needy Chappell, Microsoft general manager responsible for the development of artificial intelligence in the Azure cloud service.

The technology has allowed OpenAI to launch the popular artificial intelligence chatbot ChatGPT. The project attracted more than 1 million users within a few days of launching to all users in November 2022.

Formerly one of the founders of OpenAI Elon Musk declarationThat Microsoft is gradually “taking over” OpenAI by funding the projects of an AI-powered startup.

No time to read?
Get a summary
Previous Article

Hollywood Academy ignores Carlos Saura

Next Article

It became clear why PSG was skeptical of extending Messi’s contract.