With DGX Cloud Lepton, Nvidia proposes an AI platform that connects developers with a global network of GPU cloud providers through a centralized marketplace.
Nvidia announces DGX Cloud Lepton. This is a new platform that provides developers access to tens of thousands of GPUs from cloud providers worldwide. Participants in the platform include CoreWeave, Crusoe, Firmus, Foxconn, GMI Cloud, Lambda, Nebius, Nscale, SoftBank Corp., and Yotta Data Services, who make their GPU infrastructure available through Lepton.
The platform aims to meet the increasing demand for computing power for AI applications, including agent-based and physical AI. DGX Cloud Lepton allows developers to reserve GPU capacity by region, both on-demand and for long-term use.
All Nvidia
According to Nvidia, this enables companies to better respond to data sovereignty requirements and low latency needs. The fact that Nvidia also lowers the threshold for using its own ecosystem is a nice bonus.
The platform integrates with the existing Nvidia software stack, including NIM and NeMo microservices, Blueprints, and Cloud Functions. This combination is meant to simplify the development and deployment process of AI applications, again within its own environment. For cloud providers, DGX Cloud Lepton offers management features such as real-time diagnostics of GPU performance and automated error analysis.
read also
Nvidia Showcases RTX Pro Servers: New Design for AI Workloads
DGX Cloud Lepton is built for developers. They can sign up for early access to the platform, which is ultimately intended to grow into a central hub for global GPU computing capacity.
Exemplar Cloud
Parallel to the launch of Lepton, Nvidia also introduces its Exemplar Cloud initiative. With this, the company aims to support cloud partners in optimizing the security, performance, and reliability of their AI services. This is again done using Nvidia’s reference hardware, software, and operational tools, as the company further strengthens its grip on the GPU compute ecosystem at this edition of Computex.