NVIDIA launches DGX Cloud Lepton to link global GPU networks

NVIDIA has unveiled DGX Cloud Lepton, a platform aimed at linking developers with a worldwide network of cloud GPU providers through an AI compute marketplace.

The DGX Cloud Lepton platform brings together GPU resources from companies such as CoreWeave, Crusoe, Firmus, Foxconn, GMI Cloud, Lambda, Nebius, Nscale, SoftBank Corp and Yotta Data Services. These partners are making tens of thousands of GPUs available through the platform, including those based on the NVIDIA Blackwell architecture, in order to support increasing demand for generative and physical AI applications.

Developers using the platform are able to access GPU computing power in specific geographic regions for both on-demand and long-term requirements. This is aimed at supporting operational needs, such as strategic and sovereign AI initiatives which require compliance with regional data regulations. NVIDIA has also signalled that additional cloud service providers and GPU marketplaces are expected to join the DGX Cloud Lepton marketplace in the future.

Jensen Huang, Founder and Chief Executive Officer of NVIDIA, said, "NVIDIA DGX Cloud Lepton connects our network of global GPU cloud providers with AI developers. Together with our NCPs, we're building a planetary-scale AI factory."

The DGX Cloud Lepton platform is designed to tackle a consistent challenge in the AI sector: reliable access to sufficient, high-performance GPU resources. To address this, it combines access to cloud AI services and GPU capacity across the NVIDIA ecosystem. The system integrates with existing NVIDIA software, including NIM and NeMo microservices, NVIDIA Blueprints, and NVIDIA Cloud Functions. This aims to accelerate and streamline the development and deployment lifecycle for AI applications by providing unified tooling and interfaces.

For cloud providers, the DGX Cloud Lepton platform includes management software that supplies real-time GPU health diagnostics and automation for root-cause analysis. This is intended to reduce manual intervention and lower system downtime.

The platform is introduced with multiple touted benefits. In terms of productivity and flexibility, it delivers a single experience for development, training, and inference. Developers can buy GPU capacity straight from participating cloud providers or use their own clusters, offering greater autonomy over deployment. The platform also aims to allow frictionless deployment of AI applications across multi-cloud and hybrid settings, making it easier to handle inference, testing, and training on varying workloads.

Another highlight is the agility provided to users. The ability to quickly access GPU resources in select regions is designed to help developers meet requirements for data sovereignty and support low-latency workloads. According to NVIDIA, the platform offers enterprise-level reliability, performance, and security, helping partners deliver a uniform user experience.

NVIDIA also introduced Exemplar Clouds, a programme which assists cloud partners in improving security, usability, performance and resiliency. Exemplar Clouds utilise NVIDIA's reference hardware, software, operational tools, and benchmarking suite—DGX Cloud Benchmarking—to assist partners in optimising workload performance and evaluating cost-effectiveness.

Yotta Data Services has been named as the first NVIDIA Cloud Partner in the Asia-Pacific region to participate in the Exemplar Cloud initiative.

Developers can register for early access to the DGX Cloud Lepton platform. NVIDIA stated that further details about the programme and related technology advances can be explored at its events in Taipei.