OpenAI intends to build its own data centers in the United States as part of a plan to attract 1 billion users to its services and further commercialize its technologies. According to an OpenAI representative, the company intends to have a data center at its disposal for both training and inference, reports the Financial Times.
According to a company spokesperson, “chips, data and energy” will be key to helping the company win the AI technology race and create AGI that potentially matches or exceeds the capabilities of the human brain.
OpenAI intends to build data center clusters in the Midwest and southeast of the United States. Details have not yet been disclosed, but it is confirmed that the construction of data centers is part of the company’s AI technology commercialization strategy. OpenAI was originally founded as a non-profit organization, but is now transitioning to a more traditional business corporate structure. The company has hired its first head of finance and product and is now allegedly looking at integrating advertising into its solutions, the Financial Times reports.
Image source: Caspar Camille Rubin/unsplash.com
For its current digital infrastructure, OpenAI, which raised another $6.6 billion earlier this year, relies heavily on Microsoft Azure data centers—Microsoft has invested billions of dollars in the startup. Recently, the company began looking at other cloud providers, and in June announced a partnership with Oracle, within which Oracle Cloud Infrastructure will be used to train AI models (though also through the mediation of Microsoft). The company’s demands exceed the capacity that Microsoft can provide it with.
Not long ago, OpenAI tried to convince the US government of the need to create 5 GW data centers capable of accommodating 2 million accelerators at once. The company stated that this is already a matter of national security, and even prepared a number of recommendations.