Nvidia introduced DGX Cloud Lepton, an AI platform and compute marketplace connecting developers of agentic and physical AI systems to tens of thousands of GPUs via cloud service providers. This platform, supported by Nvidia Cloud Partners, offers regional GPU access, various usage plans, and tools for multi-cloud deployment and compliance. Nvidia also launched Exemplar Clouds, an initiative with partners like Yotta Data Services, to enhance system security and reliability using Nvidia’s expertise and benchmarking tools. These developments aim to expand distributed AI computing by unifying providers and developers globally.
Nvidia today introduced Nvidia DGX Cloud Lepton, a new AI platform and compute marketplace designed to link global developers working on agentic and physical AI systems with access to tens of thousands of GPUs, available through a distributed network of cloud service providers.
Nvidia’s DGX puts a supercomputer on your desk.
To support ongoing AI development needs, Nvidia Cloud Partners (NCPs)—including CoreWeave, Crusoe, Firmus, Foxconn, GMI Cloud, Lambda, Nebius, Nscale, SoftBank Corp., and Yotta Data Services—will offer GPU capacity based on Nvidia Blackwell and other Nvidia architectures through the DGX Cloud Lepton marketplace.
This platform enables developers to access GPU compute resources in specific geographic areas, supporting both short-term and extended usage plans. The system is structured to accommodate regional and organizational requirements around data use, including those related to compliance and latency. Additional cloud providers and GPU-focused marketplaces are expected to join the platform in the future.
“Nvidia DGX Cloud Lepton brings together our network of GPU cloud providers and AI developers,” said Jensen Huang, founder and CEO of Nvidia. “With our partners, we are expanding infrastructure to support a wide range of AI projects across the globe.”
DGX Cloud Lepton addresses the challenge of accessing dependable GPU compute resources by providing a single entry point to Nvidia’s wider compute framework. The platform connects to Nvidia’s broader software ecosystem, including NIM and NeMo microservices, Cloud Functions, and AI Blueprints, offering a consistent environment for creating and scaling AI systems.
For cloud infrastructure providers, DGX Cloud Lepton includes management tools for real-time monitoring of GPU status and automated fault diagnostics. This setup is intended to reduce the need for manual checks and to help minimize service interruptions.
The platform provides several core capabilities:
Unified development experience: DGX Cloud Lepton allows developers to use a consistent interface for training, inference, and deployment. GPU access can be acquired directly through participating cloud providers, or developers may use their own infrastructure alongside the marketplace.
Multi-cloud and hybrid deployment: The platform supports application deployment across multiple cloud environments with a reduced need for custom operational adjustments. Tools for inference, testing, and training are integrated to streamline this process.
Regional access and compliance: Developers can choose to run workloads in specific locations to meet legal and operational requirements related to data residency or latency. This flexibility supports a variety of industry and government use cases.
Stable performance expectations: The platform is designed to offer cloud providers a reference standard of performance and operational behavior. This consistency allows organizations to plan and manage their use of resources with greater predictability.
In parallel with the launch of DGX Cloud Lepton, Nvidia also announced Nvidia Exemplar Clouds. This new initiative supports NCPs in refining system security, usability, and reliability. It draws on Nvidia’s existing infrastructure knowledge, validated hardware configurations, and operational tooling.
Exemplar Clouds make use of Nvidia DGX Cloud Benchmarking, a toolkit that assists partners in measuring and optimizing the performance of workloads running on AI compute systems. The benchmarking suite also enables cost-to-performance evaluations, allowing partners to balance resource investment with desired outcomes.
Yotta Data Services has joined as the first NCP in the Asia-Pacific region to participate in the Exemplar Cloud initiative. Through this collaboration, Yotta aims to align its offerings with Nvidia’s operational guidelines and resource configuration recommendations.
Together, these developments signal an expansion of Nvidia’s approach to distributed AI computing. By linking cloud providers and developers through unified platforms and common tooling, the company is supporting the ongoing integration of AI across sectors that require scalable, regionally adaptable compute capacity.
LIKE WHAT YOU’RE READING? INTRODUCE US TO YOUR FRIENDS AND COLLEAGUES.