Nvidia Takes Aim at Hyperscalers with Data Center Bet

Nvidia has announced it is investing $1.5 billion in a data center campus in Ohio, to be built and operated by SB Energy. The campus will have an initial capacity of 4.25 gigawatts of IT power, with the option to expand to 8 gigawatts.

This investment marks a significant shift for Nvidia, which has been expanding its business beyond just selling graphics processing units (GPUs). Instead, it is now taking on a larger role in the AI infrastructure ecosystem by providing compute power and leasing capacity to tenants like OpenAI.

Nvidia's move into data centers is driven by the increasing complexity of AI infrastructure. While GPUs can be manufactured quickly, securing land, power, and physical space for data centers is a much more time-consuming process. Nvidia is now taking on this challenge directly, positioning itself as a provider of both compute hardware and infrastructure.

This shift in strategy has significant implications for the broader industry, including AWS, Azure, and Google Cloud. While these hyperscalers have been trying to reduce their dependence on Nvidia by developing custom silicon, Nvidia is now going around them by leasing capacity directly to tenants like OpenAI.