Nvidia recently launched the DGX Cloud Lepton, a new AI-focused cloud software platform. This service acts as a marketplace where AI factories can rent out their computing resources to developers needing high-performance computing capabilities across global locations.
During the announcement, Alexis Bjorlin, Nvidia’s Vice President of DGX Cloud, likened Lepton to ridesharing apps like Uber or Lyft, but instead of connecting riders with drivers, it links developers to GPU computing resources. He emphasized that DGX Cloud Lepton allows developers to access GPU compute capacity globally, enhancing availability across various cloud services and geographical regions.
Although currently in its early access phase, several companies, including CoreWeave, Crusoe, and Foxconn, have committed to making "tens of thousands of GPUs" available on the platform. This initiative enables developers to utilize GPU capacity for both transient and extended computing needs, aligning with strategic and sovereign AI operational requirements. Nvidia anticipates that prominent cloud service providers and GPU marketplaces will join the DGX Cloud Lepton ecosystem.
The platform integrates Nvidia’s AI software stack—comprising tools such as NIM, NeMo microservices, Blueprints, and Cloud Functions—to expedite and streamline the creation and deployment of AI applications. DGX Cloud Lepton offers a cohesive experience throughout development stages, covering training and inference, along with facilitating the deployment of AI applications across multi-cloud and hybrid environments efficiently.
Nvidia’s acquisition of Lepton AI just last month enabled this rapid rollout. This earlier venture provided GPU rental services by leasing computing resources from Cloud Service Providers (CSPs) and delivering them as an AI-as-a-Service model to smaller clients, also offering features like autoscaling and error management.
For more updates on Nvidia’s initiatives, you can check related content:
