Site icon Baltimore Chronicle

Microsoft will build Stargate data center in the desert for OpenAI for $100 billion

According to several media sources, Microsoft will finance the construction of the world's largest data center called Stargate for OpenAI. The project is estimated at $100 billion, will take about six years to implement and will include a cluster of supercomputers with artificial intelligence.

As part of the Stargate project, every supercomputer The cluster will include “millions of specialized server chips” to support OpenAI AI systems.

Experts believe that this is Microsoft's most ambitious project, “which will be 100 times more expensive than some of today's largest data centers.”

Microsoft and OpenAI plan to launch the first Stargate supercomputer in 2026, and in 2028, a series of similar installations will be added to it. Microsoft management expects Stargate to be launched at full capacity before 2030.

Project details: what determines the launch date of the supercomputer

According to media reports, the implementation of the Stargate project depends on whether OpenAI can present its next-generation AI model GPT-5 within a certain time frame. The company plans to launch GPT-5 in early 2025.

In March of this year, the media wrote that OpenAI has been actively testing the next generation of an artificial intelligence model called GPT-5 in a closed format for some time. According to testers, this model is significantly better than GPT-4. New options are available there, such as an intelligent agent that can independently carry out human tasks over long periods of time, as well as various systems for autonomously performing tasks.

Earlier, OpenAI head Sam Altman said that the developers are training the AI ​​model of the fifth generation not only on open information and combinations of publicly available data sets on the Internet, but also on the internal databases of many large IT companies that cooperate with OpenAI.

How OpenAI is developing

In early March 2024, researchers determined that OpenAI was working on the GPT-4.5 Turbo neural network, which has a context length window of 256 KB of tokens, double the current 128 KB of GPT-4 Turbo.

In March 2023 OpenAI introduced the GPT-4 image and text interpretation AI model, which the company called “the latest milestone in its efforts to expand the scope of deep learning.”

In early 2023, Microsoft and OpenAI announced they were expanding their partnership and “ Microsoft's multi-year, multi-billion dollar investment in OpenAI.

minfin.com.ua

Exit mobile version