Case Study: Cohere achieves 3x faster AI training with CoreWeave

Cohere achieves 3x faster AI training with CoreWeave

Cohere, an AI native company building enterprise-ready agentic AI systems, faced the challenge of needing a massively scalable compute and high-performance storage foundation to support the training of its North AI platform. They required a cloud partner that could deliver cutting-edge performance and operational reliability to run next-generation workloads at scale.

CoreWeave deployed one of the industry’s first NVIDIA GB200 NVL72 clusters for Cohere, combined with their high-performance AI Object Storage. This solution provided Cohere with early access to new hardware and proactive operational management. As a result, Cohere achieved 3x faster training performance for its large-scale models and realized storage throughput of up to 7 GB/s per GPU, accelerating their time to market for next-generation AI.

Cohere