Nvidia launches GPU cloud marketplace, announces AI campuses in Taiwan, France, and Middle East

Summary

Nvidia launched DGX Cloud, a GPU marketplace that makes its AI chips directly available to developers across multiple cloud platforms. The service expands on what Dylan Patel identified as one of Nvidia's structural advantages over AMD: the ability to seed researchers and universities with GPU access before they commit to a platform. Nvidia historically distributed what Patel calls "goldplated racks of GPUs" to research labs, letting developers train on the hardware and become fluent in CUDA. That familiarity created switching costs without formal contracts.

DGX Cloud uses a broker model instead of direct supply. Private companies already offer similar services, though the specific competitive dynamics remain unclear.

The move reflects Nvidia's belief that cloud distribution is now critical to maintaining developer mindshare as AI infrastructure commoditizes. By embedding its chips into the cloud marketplace layer, Nvidia aims to maintain switching costs at the software level through the CUDA ecosystem and developer familiarity, even as cloud providers proliferate and GPU availability becomes less scarce.

Jensen Huang also announced AI research campuses in Taiwan, France, and the Middle East. The announcement signals Nvidia's intent to become a full-stack infrastructure player rather than a chip supplier alone. No financial figures, timelines, or campus details were disclosed.