Case Study: Cartesia ships sub-120ms real-time voice AI with Crusoe Cloud

Cartesia ships sub-120ms voice AI with Crusoe

Cartesia, a company building next-generation real-time multimodal AI, faced the challenge of training its novel state space model architecture. They required a cloud infrastructure partner that could provide strong price-to-performance for sustained training, highly responsive support, and operational reliability at scale. They partnered with Crusoe to address these needs.

Crusoe implemented a solution using NVIDIA Hopper GPU clusters on Crusoe Cloud and brought in a subject matter expert to build a custom SLURM cluster specifically for Cartesia. This partnership enabled Cartesia to ship its Sonic text-to-speech model with a sub-120ms latency, the fastest in its class at launch. As a result of Crusoe's support, Cartesia tripled its infrastructure footprint and expanded its storage instances, solidifying a continued partnership for future innovation.

Cartesia