CoreWeave (NASDAQ:CRWV) saw its stock rise 4% on Wednesday after the company disclosed a multi-year strategic partnership with Perplexity to host the latter’s AI inference workloads on CoreWeave’s cloud platform.
Under the deal, Perplexity will run its next-generation inference workloads on CoreWeave infrastructure using dedicated NVIDIA GB200 NVL72-powered clusters. The arrangement is intended to support growth in Perplexity’s Sonar and Search API ecosystem as usage increases.
As part of the agreement, CoreWeave will implement Perplexity Enterprise Max across its organization. The enterprise product will enable CoreWeave employees to search both the open web and internal knowledge bases, perform research, visualize data, and access AI models from within a single platform.
Commenting on the partnership, Max Hjelm, senior vice president of revenue at CoreWeave, said: "We’re proud to partner with Perplexity as they scale their inference workloads on CoreWeave’s AI cloud," said Max Hjelm, senior vice president of revenue at CoreWeave. "AI applications running in production require more than just access to raw infrastructure - they require best-in-class performance and reliability as well as a cloud platform designed end-to-end for AI that simplifies compute operations."
Perplexity has already initiated the initial deployment phase by running inference workloads with CoreWeave Kubernetes Service. In addition, Perplexity is using W&B Models to train, fine-tune, and manage models as they move from experimentation to production.
The companies described the collaboration as reflecting Perplexity’s multi-cloud strategy and positioning CoreWeave as a specialized AI cloud provider for organizations operating advanced AI systems in production environments.
CoreWeave also highlights its performance credentials: it holds the only AI cloud to earn a top Platinum ranking in both SemiAnalysis ClusterMAX 1.0 and 2.0, assessments that evaluate AI cloud performance, efficiency, and reliability.
Context and market effect
The announcement combined an immediate market reaction in CoreWeave’s share price with operational actions including starting inference runs on Kubernetes and integrating enterprise software internally. The technical elements named in the deal - dedicated NVIDIA GB200 NVL72-powered clusters and W&B Models for model lifecycle management - underline the partnership’s focus on production-grade inference capacity and model operations.