Nvidia’s vision for data centers may be a false prophecy

Just two years ago, Nvidia was primarily known for its video game hardware. Today, it is the world’s most valuable company, responsible for the chips that power the largest and most complex generative AI models. This rapid transformation has enabled Nvidia to set the agenda for much of the conversation on the digital infrastructure requirements for large-scale AI deployments.

Nvidia’s vision for the future of AI involves highly performant, power-hungry GPUs in very dense server configurations — the kind of hardware that can only be delivered by Nvidia. This type of infrastructure requires increased amounts of power and cooling typically associated with supercomputing, which means major changes in data center design and operation.