Thunder Compute

Thunder Compute is a cloud platform designed to make high-performance NVIDIA H100 GPUs directly accessible from familiar development environments. Its primary purpose is to let engineers, researchers, and data scientists spin up H100 instances quickly, run intensive workloads, and manage compute resources without complex cloud setups or long-term contracts. By integrating directly into VS Code and offering multiple access methods, Thunder Compute streamlines the end-to-end workflow for AI training, inference, and other GPU-heavy tasks while significantly reducing costs compared to traditional cloud providers.

Key features include a VS Code extension that allows users to provision and control H100 GPUs from within their editor, enabling seamless code–to–compute workflows. A command-line interface (CLI) is available for scripting, automation, and integration into existing DevOps pipelines. Browser-based access provides a way to manage and monitor GPU instances without installing additional tools, while expandable storage supports larger datasets and model checkpoints. Pricing starts at $1.38 per hour for NVIDIA H100 GPUs, with no contract requirements and the potential to save up to 80% compared to major cloud platforms such as AWS.