NVIDIA has introduced DGX Cloud Lepton, a new AI platform and compute marketplace designed to connect developers with a global network of GPU resources. Announced during COMPUTEX on 18 May, the service brings together infrastructure from leading NVIDIA Cloud Partners (NCPs), offering access to tens of thousands of GPUs to support the development of agentic and physical AI applications.
Participating cloud providers include CoreWeave, Crusoe, Firmus, Foxconn, GMI Cloud, Lambda, Nebius, Nscale, SoftBank Corp., and Yotta Data Services. These partners will offer NVIDIA’s latest GPU architectures, including the Blackwell series, through the DGX Cloud Lepton marketplace. The platform is built to support both on-demand and long-term compute needs, allowing developers to deploy workloads in specific regions to meet strategic and sovereign AI requirements.
“NVIDIA DGX Cloud Lepton connects our network of global GPU cloud providers with AI developers,” said Jensen Huang, founder and CEO of NVIDIA. “Together with our NCPs, we’re building a planetary-scale AI factory.”
Enhancing access and efficiency for developers
DGX Cloud Lepton addresses a core challenge in the AI development landscape: securing consistent, high-performance GPU access. By bringing together the offerings of various cloud partners into a centralised marketplace, NVIDIA aims to create a seamless, scalable solution for AI teams across the world. The platform integrates tightly with the NVIDIA software stack, including services such as NIM, NeMo microservices, Blueprints and Cloud Functions.
These integrations are designed to reduce friction in the development pipeline, allowing teams to accelerate everything from model training to deployment. Developers can either purchase GPU capacity directly from cloud providers listed on the marketplace or choose to integrate their own compute clusters, giving them flexibility and greater control over their infrastructure strategy.
From a performance perspective, DGX Cloud Lepton promises predictability and reliability. The platform provides enterprise-grade compute capabilities with secure, consistent performance, making it a viable option for organisations that require high levels of uptime and data compliance.
Operational support for cloud providers
In addition to serving developers, DGX Cloud Lepton also offers tools for cloud infrastructure providers. Management software included with the platform delivers real-time GPU health diagnostics and automates root-cause analysis. These features are aimed at reducing downtime and streamlining troubleshooting, ultimately lowering operational overhead for providers.
The platform’s design also allows for deployment across multi-cloud and hybrid environments. By supporting regional resource allocation, DGX Cloud Lepton helps organisations comply with data sovereignty regulations and address latency-sensitive use cases. The unified platform experience helps developers move from development to inference more efficiently, without needing to manage multiple fragmented systems.
NVIDIA launches Exemplar Clouds to raise performance standards
Alongside DGX Cloud Lepton, NVIDIA announced its Exemplar Clouds initiative to help cloud partners optimise AI performance and infrastructure. These model environments use NVIDIA’s benchmark tools, reference hardware, and operational practices to improve usability, performance, and resilience across partner platforms.
Exemplar Clouds are powered by NVIDIA DGX Cloud Benchmarking — a suite of tools and guidelines aimed at helping providers understand and enhance workload performance while maintaining cost efficiency. These tools offer a standardised method to assess and improve service delivery for AI workloads.
Yotta Data Services has become the first cloud partner in the Asia-Pacific region to join the Exemplar Cloud programme. By adopting NVIDIA’s reference practices, Yotta aims to offer more robust AI services to clients in the region and better align with global performance expectations.