Lambda has announced the general availability of multi-node NVIDIA HGX B200-accelerated clusters accessible on-demand through its 1-Click Clusters. This service allows Artificial Intelligence (AI) engineers and researchers to train and infer using advanced NVIDIA Blackwell GPUs without the need for long-term contracts. The on-demand model is designed to meet the increasing demand for computational power within the AI sector.

By leveraging the capabilities of NVIDIA Blackwell GPUs with self-serve access, Lambda aims to streamline the workloads of AI teams, facilitating faster innovation and simplifying infrastructure management. Clients can now instantly access between 16 to 512 clusters connected via NVIDIA Quantum-2 InfiniBand, which supports training and fine-tuning on NVIDIA’s latest hardware.

Stephen Balaban, CEO and co-founder of Lambda, emphasized the importance of readily available infrastructure for AI development, stating, “AI innovation requires immediate access to cutting-edge infrastructure that scales with the needs of AI teams.” He highlighted the initiative's goal of effectively supporting AI developers.

Dhanji R. Prasanna, CTO of Block, indicated a commitment to exploring new solutions through these clusters, emphasizing an open-source approach in sharing knowledge and results. The Lambda 1-Click Clusters are tailored to the demands of modern AI engineers, providing a flexible environment to experiment with new technologies.

Dave Salvator, Director of Accelerated Computing Products at NVIDIA, noted that the partnership aims to provide fast, efficient access to advanced infrastructure, which is critical for organizations looking to advance their AI initiatives. Interested parties can explore the option to launch NVIDIA HGX B200 clusters on-demand directly through Lambda’s platform.