Microsoft and OpenAI plan to build a data center with millions of servers

Microsoftand OpenAI A breakthrough is reportedly being plannedData CenterThe project will include an artificial intelligence called StargatesuperThe goal of the project, The Information’s Anissa Gardizy and Amir Efrati report, is to reduce the two companies’ reliance on Nvidia, something many tech giants dabbling in artificial intelligence are increasingly trying to do.

Microsoft and OpenAI plan to build a data center with millions of servers

Microsoft is said to be investing more than $100 billion in the project and has reportedly set a launch date of 2028. Microsoft's plan with OpenAI reportedly includes five phases, with Stargate being the fifth and most ambitious.

While the cost of the project is based on “people familiar with the plans” (which The Information says includes “people who have discussed it with OpenAI CEO Sam Altman and people who have reviewed Microsoft’s preliminary cost estimates”), Microsoft and OpenAI have yet to comment on the specifics of the project.

The new data center project is expected to push the boundaries of AI capabilities and could exceed $115 billion in spending. That’s more than three times the amount Microsoft spent on capital expenditures for servers and equipment last year. Microsoft is currently developing a smaller fourth phase for OpenAI, according to The Information.supercomputer, which is expected to be launched around 2026.

The Next Platform further explains, “News about Microsoft’s plans to build a system rumored to be called ‘Stargate’ to support the computational needs of its large language model partner OpenAI.FirstThe thing is that people are talking about data centers, notsuperThat’s because data centers—perhaps multiple data centers in a single region, potentially housing as many as a million XPU computing devices—will becomesupercomputer."

The Next Platform also says that if Stargate is indeed built, it will be based on a future generation of Cobalt Arm server processors and Maia XPUs, with Ethernet scaling to hundreds of thousands to a million XPUs on a single machine, andabsoluteIt won't be based on Nvidia's GPU and interconnect technology, if these rumors are to be believed.

statement:The content is collected from various media platforms such as public websites. If the included content infringes on your rights, please contact us by email and we will deal with it as soon as possible.
Information

In addition to singing, can AI also perform for you? Another AI company pushes virtual humans to a new level

2024-4-15 10:23:36

Information

The UK is drafting AI regulations to ensure its development is under control

2024-4-16 9:31:58

Search