Crusoe and Perplexity announce multi-year AI infrastructure deal

Crusoe Cloud has entered into a multi-year agreement with Perplexity to supply AI infrastructure for the company's workloads. Perplexity will conduct all stages of its model lifecycle, from training frontier models to deploying them, on Crusoe's platform. This includes utilizing NVIDIA GB300 NVL72 and NVIDIA InfiniBand Solutions for training, and employing Crusoe's Managed Inference service for deployment.

Crusoe's platform will also offer web and internal knowledge search, multi-step research, data analysis, and access to frontier AI models to Perplexity's 1,800 employees. Both companies emphasize the importance of infrastructure that can keep up with the rapid pace of AI development and scale accordingly. Crusoe's Managed Inference service is specifically designed for production-grade inference, focusing on performance and cost efficiency for popular models.

For Perplexity, which aims to revolutionize how people find answers, Crusoe serves as a crucial partner, from the initial stages of model development to the final output. The partnership enables Perplexity to efficiently train, fine-tune, and serve frontier models using NVIDIA's cutting-edge hardware.