Nvidia scales down DGX Cloud ambitions amid hyperscale tensions
Nvidia is scaling down its vision for DGX Cloud, a project once touted as a pathway into the broader cloud market but now largely redirected to serve its own research teams. The move comes after the service struggled to attract widespread adoption while creating unease among the very hyperscale partners that supply the infrastructure it relies on.
Launched as a cloud platform built on top of major providers like AWS and Google, DGX Cloud aimed to rent GPU power to enterprise customers who wanted ready access to Nvidia’s hardware for AI development. The company even projected that such efforts could eventually generate more than $150 billion in revenue. Yet the reality has been more complicated. Customers often found it cheaper to work directly with the big cloud providers rather than through Nvidia, leaving DGX Cloud with limited traction.
The Information reported that Nvidia has stopped onboarding new customers and no longer highlights DGX Cloud in its earnings disclosures. Instead, the company has committed to using the platform internally, with executives describing it as a fully utilized environment for AI research and a collaborative space for optimizing compute stacks with partners.
“DGX Cloud is a great success as a cloud for Nvidia’s own AI research and development,” said Alexis Black Bjorlin, who leads the initiative, disputing suggestions that the project has failed. She emphasized that demand remains strong inside Nvidia and that the company continues to expand its capacity.
Earlier this year, Nvidia introduced DGX Cloud Lepton, a marketplace designed to connect GPUs from providers including CoreWeave, SoftBank, and Yotta Data Services. Even though the action allowed new ways to get access, it still stirred up conflicts with those in the business of infrastructure who were concerned that Nvidia was going into their client base too much.
The change represents the far side of the fence for Nvidia. Firstly, the company is still the main provider of GPUs that drive AI advancements. On the other, its attempts to play the role of cloud operator risk straining ties with the companies that make up much of its downstream demand. For now, DGX Cloud’s future appears less about becoming a stand-alone revenue engine and more about supporting Nvidia’s internal development and partner ecosystem.

