Nvidia acquires Lepton AI to build global GPU compute marketplace

In April 2025, Nvidia made a notable move in the tech industry by acquiring Lepton AI, a Chinese startup specialized in GPU cloud services. Founded in 2023, Lepton AI had focused on providing rental GPU computing resources aggregated from various infrastructures and cloud providers. The acquisition aimed to bolster Nvidia's offerings by integrating Lepton AI's innovative technology into its existing portfolio. Following the acquisition, the founders of Lepton AI, Yangqing Jia and Junjie Bai, joined Nvidia to continue their work on the platform. After the acquisition, Nvidia rebranded Lepton AI as DGX Cloud Lepton, with an official relaunch occurring in June 2025. The new service promises to deliver a unified AI platform and compute marketplace that connects developers to a vast network of GPUs through partnerships with global cloud providers. Nvidia designed DGX Cloud Lepton to aggregate GPU capacity from major industry players like AWS, CoreWeave, and Lambda via a single software interface. This feature aims to simplify access for developers to GPU computing resources, regardless of the geographical location of the servers. DGX Cloud Lepton supports three primary workflows, which are essential for developers working on AI applications. First, it offers Dev Pods—interactive environments for prototyping and experimentation using tools like Jupyter notebooks and VS Code. Second, it provides Inference Endpoints that allow users to deploy models as scalable endpoints. Lastly, the service features a suite of operational capabilities such as real-time monitoring, auto-scaling, custom workspaces, and compliance with security requirements. Developers have the flexibility to choose their data region, which helps meet data locality and sovereignty needs. At the recent GTC event in Paris, Nvidia announced partnerships with leading cloud providers in Europe to further address local developers' data sovereignty concerns. Additionally, Nvidia’s collaboration with Hugging Face aims to facilitate training clusters as a service for AI developers. Nvidia is also working with European venture capital firms to support eligible portfolio firms by providing GPU capacity credits and technical assistance. With most of these services still in the preview stage, developers are encouraged to apply for early access through Nvidia's platform. Ultimately, by aligning with existing cloud infrastructures instead of creating a competing service, Nvidia is showing an innovative approach to meeting the growing demand for GPU resources among developers worldwide.