Crusoe to Run Perplexity Model Training and Inference

Crusoe and Perplexity AI announced a multiyear partnership covering the full model lifecycle. Perplexity will train frontier models on Crusoe Cloud and serve them in production through Crusoe's Managed Inference service.

Training will run on dedicated clusters powered by NVIDIA GB300 NVL72 systems and InfiniBand networking. Crusoe's Managed Inference service will serve Perplexity's models in production.

Crusoe will also provide Perplexity Enterprise Pro and Max to its 1,800 employees. The tools support web and internal knowledge search, multi step research, data analysis and access to frontier models. Perplexity co-founder and chief executive Aravind Srinivas said running AI at Perplexity's scale means every millisecond of latency is felt by users.