Report: Microsoft to construct ‘Stargate’ supercomputer with tens of millions of chips for OpenAI

Microsoft Corp. plans to construct a supercomputer with tens of millions of processors to support OpenAI’s research, The Information reported today.

The system will likely be geared toward running artificial intelligence workloads. Reportedly codenamed Stargate by Microsoft executives, the supercomputer is believed to be a part of a broader initiative that can be set to see the corporate construct several other AI clusters. The project is predicted to cost as much as $100 billion.

OpenAI already uses Microsoft infrastructure to coach its AI models. In 2020, the cloud computing and software giant disclosed that it had built an Azure-hosted supercomputer with 10,000 graphics cards to support OpenAI’s work. Based on the businesses, the system ranked as considered one of the world’s five fastest supercomputers on the time of its launch.

Last March, Microsoft provided an update about its infrastructure collaboration with OpenAI. The corporate detailed that the unique 10,000-GPU supercomputer it had built for the AI developer has since been upgraded to incorporate tens of 1000’s of A100 chips. Microsoft executive Scott Guthrie stated that the system’s cost was “probably larger” than several hundred million dollars.

Today’s report from The Information indicates more upgrades are within the works. Based on the publication’s sources, Microsoft plans to construct several additional AI infrastructure installations through 2030. The plan is reportedly divided into five phases, with Microsoft and OpenAI currently believed to be in the midst of the third.

The fourth phase is predicted to involve the development of a brand new supercomputer that can launch “around” 2026. The fifth phase, in turn, will center on the system that Microsoft executives check with as Stargate internally. The supercomputer and its tens of millions of chips are expected to develop into operational as early as 2028.

As of last March, Microsoft mainly used Nvidia Corp. graphics cards to power the supercomputing infrastructure it provides to OpenAI. That would potentially change going forward. In November, the corporate detailed an internally developed AI accelerator called Azure Maia that features 105 billion transistors.

It’s possible OpenAI won’t be the only real user of Stargate and the opposite recent AI clusters Microsoft is predicted to deploy. Last yr, the latter company detailed that it has used supercomputing infrastructure built for OpenAI to coach its own AI models. Microsoft has also made the hardware available to cloud customers, an approach it could take with Stargate as well to more quickly recoup the system’s likely steep cost.

Provided that it’s expected to come back online in 2028 on the earliest, Stargate likely won’t be used to coach the successor to OpenAI’s GPT-4. Last week, Insider reported that GPT-5 is ready to develop into available in a number of months. It’s believed that OpenAI already has a working prototype of the model with “materially higher” capabilities than GPT-4. 

Photo: Microsoft