Microsoft and OpenAI are reportedly planning a groundbreaking data center project, including an AI supercomputer called “Stargate.”
A report by Anissa Gardizy and Amir Efrati in The information claims that the aim of the project, which would be funded by Microsoft to the tune of over $100 billion and which reportedly has a launch date of 2028, is to reduce the two companies’ dependence on Nvidia, something that many of the technology companies giants involved in AI are increasingly looking to do so.
Microsoft and OpenAI’s plan reportedly includes five phases, with Stargate being the fifth and most ambitious phase.
The data center becomes the supercomputer
The cost of the project is attributed to age-old “sources familiar with the plans” (the Information says this is “a person who spoke to OpenAI CEO Sam Altman about it and a person who provided some of the initial cost estimates from Microsoft’), but neither Microsoft nor OpenAI have yet commented on the details of the project.
The new data center project is expected to push the boundaries of AI capabilities and could potentially exceed $115 billion in expenditures. This is more than triple the amount Microsoft spent on capital expenditures on servers and equipment last year. Microsoft is currently working on a smaller, fourth-stage supercomputer for OpenAI, expected to launch around 2026. The information claims.
To shed more light on the report, The next platform says: “The first thing you’ll notice about the rumored ‘Stargate’ system that Microsoft plans to build to support the computational needs of its major language model partner, OpenAI, is that the people speaking – reportedly OpenAI – CEO Sam Altman – are talking about a data center, not a supercomputer. And that’s because the data center – and perhaps multiple data centers within a region with perhaps 1 million XPU computing devices – will be the supercomputer.”
The next platform also says that if Stargate becomes a reality, it will be “based on future generations of Cobalt Arm server processors and Maia XPUs, with Ethernet scaling to hundreds of thousands to 1 million XPUs in a single machine,” and that is certainly the case. will not be based on Nvidia GPUs and interconnects, which seems like a safe bet if the rumors are to be believed.