Microsoft and OpenAI are planning to establish a data center with a cost of no less than $100 billion, according to The Information, which cites sources close to the companies’ leadership.
Alongside algorithm improvements, hardware, particularly specialized chips, plays a crucial role in the advancement of the AI industry. The rapid adoption of generative neural networks has led to a swift increase in demand for specialized data centers.
The project includes the creation of the Stargate supercomputer, scheduled for launch between 2028 and 2030. It will be the largest in a series that the companies plan to build over the next six years.
The development is currently in the middle of the third phase out of five. A significant portion of the expenses for the next two stages is associated with acquiring the necessary AI chips.
According to the publication, the new project will be designed to work with processors from various suppliers.
The fourth phase, planned for 2026, includes the creation of a supercomputer for OpenAI. The fifth phase involves constructing the largest supercomputer on the market, Stargate, with millions of AI chips. It is expected to be completed by the end of the decade.
“We are always thinking about the next generation of infrastructure innovations needed to expand the capabilities of neural networks,” a Microsoft representative told Reuters.
The Information suggests that expenses could exceed $115 billion, which is three times more than Microsoft’s spending on servers, buildings, and other equipment in 2023.
In April of the same year, the publication reported on the company’s work on its own AI chips for training large language models.
Back in February 2024, OpenAI discussed raising $7 trillion for the development of AI chip production, according to WSJ.