OpenAI and Microsoft are reportedly collaborating on an ambitious project to build a supercomputer named “Stargate,” with an estimated cost of up to $100 billion.
This initiative is part of a broader five-phase plan, with Stargate representing the fifth phase, expected to launch as early as 2028.[the_ad id=”9198″]
The Stargate supercomputer is envisioned to be a U.S.-based data center housing millions of specialized AI chips, designed to massively accelerate the development of OpenAI’s AI models.
The project aims to be the largest of a series of installations that the companies plan to build over the next six years.
Microsoft is likely to finance the project, which is projected to cost about 100 times more than some of the largest data centers today. The supercomputer would require several gigawatts of power, equivalent to at least several large data centers currently in operation.[the_ad id=”9198″]
A significant portion of the project’s cost would be allocated to procuring the chips, and ensuring an adequate power supply would also be a challenge.
The partnership between Microsoft and OpenAI has been under scrutiny from E.U. and U.K. regulators, and Tesla CEO Elon Musk, an OpenAI co-founder and former executive, has sued over the partnership.
Despite these challenges, the project reflects the companies’ long-term vision for AI development, with OpenAI CEO Sam Altman having previously expressed concerns about a shortage of chips for AI development.[the_ad id=”9199″]
READ ALSO: Meta adds AI to Glasses
Microsoft and OpenAI are also working on a smaller supercomputer, known as “Phase 4,” which is expected to be operational in 2026 and could cost $10 billion.
The decision to move forward with the Stargate project will depend on whether OpenAI can significantly improve the capabilities of its AI.
The demand for AI chips has surged, with companies like Nvidia currently dominating the market. OpenAI has indicated a desire to build a global infrastructure of chip factories to address this issue.[the_ad id=”9199″]
Microsoft has also developed its own AI accelerator, Azure Maia, and may consider using alternative power sources like nuclear energy for the project.
The exact location of the Stargate supercomputer is still to be determined, and it may be built in a single data center or several nearby centers. The project is still in the planning stages, and many details, including the choice of chips and networking solutions, are yet to be finalized.