TLDR
- CoreWeave shares rose nearly 6% in premarket trading after the company announced a multi-year infrastructure agreement with Perplexity AI.
- Perplexity selected CoreWeave as a backend cloud partner to run inference workloads on NVIDIA GB200 NVL72 GPU clusters.
- The infrastructure will support Perplexity’s Sonar and Search API products as the company expands its AI search services.
- Perplexity has already begun deploying workloads through CoreWeave’s Kubernetes service as part of its multi-cloud strategy.
- CoreWeave will also adopt Perplexity Enterprise Max internally to provide employees with AI-powered research and web search tools.
CoreWeave announced a multi-year agreement to run inference workloads for Perplexity AI, an emerging AI-driven search company backed by Jeff Bezos and Nvidia. The announcement pushed CoreWeave stock higher during Wednesday’s premarket trading. The companies confirmed that CoreWeave will provide specialized GPU cloud infrastructure for Perplexity’s expanding search and API services.
CoreWeave Shares Climb After Perplexity Infrastructure Partnership
CoreWeave shares gained nearly six percent in premarket trading following the infrastructure agreement with Perplexity. The partnership assigns CoreWeave as a backend cloud provider for the AI search platform.
The companies confirmed that Perplexity will deploy inference operations on NVIDIA GB200 NVL72 clusters managed by CoreWeave. These clusters support next-generation AI workloads that demand high computing performance.
Perplexity plans to run production inference tasks on dedicated GPU infrastructure across CoreWeave’s cloud environment. The system will support growing demand from its Sonar and Search API products.
Max Hjelm, CoreWeave senior vice president of revenue, described the requirements for modern AI deployments. He said, “AI applications running in production require more than raw infrastructure and demand performance, reliability, and a cloud platform designed for AI.”
He added that CoreWeave aims to simplify compute operations for AI developers building production systems. The company designed its platform specifically to support large-scale machine learning operations.
Perplexity confirmed that inference performance plays a central role in the user experience of AI search services. Fast response times and stable latency determine how quickly the platform delivers results.
The agreement positions CoreWeave infrastructure as a foundation for future product expansion across Perplexity’s services. The companies confirmed that inference workloads have already begun running on CoreWeave systems.
AI Infrastructure Expansion Supports Perplexity Growth
Perplexity executives pointed to CoreWeave’s technical capabilities when explaining the partnership decision. Dmitry Shevelenko, chief business officer at Perplexity, praised the provider’s engineering support.
He said the company valued CoreWeave’s technical expertise and its approach to working with AI-focused companies. Shevelenko stated the partnership would help accelerate infrastructure efficiency and product development.
Perplexity confirmed it already runs workloads through CoreWeave’s Kubernetes service for containerized computing tasks. The company also uses W&B Models for training and fine-tuning models within a broader multi-cloud strategy.
The deployment supports the scaling of AI search systems that process large volumes of user queries. Inference workloads execute trained models to generate answers, recommendations, and search results.
CoreWeave said specialized GPU cloud platforms have become central for AI companies handling growing compute requirements. The provider reported strong results in MLPerf benchmark evaluations for machine learning performance.
The company also holds platinum rankings in SemiAnalysis ClusterMAX performance and reliability assessments. These rankings measure computing efficiency and operational stability across large GPU clusters.





