CoreWeave announced a multi-year strategic partnership with Perplexity, the AI-powered search startup, to host its inference workloads on dedicated NVIDIA GB200 NVL72-powered clusters. The deal sent CoreWeave shares up 4% in premarket trading. Perplexity will run its Sonar platform and Search API ecosystem on CoreWeave Cloud, using Kubernetes Service and W&B Models for model refinement.

CoreWeave CEO Mike Intrator said the partnership "reflects a wider mix of emerging AI leaders adopting CoreWeave platform — they choose us for unified AI cloud, not just capacity." Perplexity CBO Dmitry Shevelenko called CoreWeave "essential to optimize our infrastructure and models to provide users with the strongest AI tools on the market."

As part of the agreement, CoreWeave will also deploy Perplexity Enterprise Max internally for web search, knowledge search, and data visualization. The deal underscores the growing demand for specialized AI cloud infrastructure as companies move from training to production-scale inference.