Perplexity Taps CoreWeave Cloud for NVIDIA GB200-Powered AI Inference
CoreWeave has secured a multi-year partnership with Perplexity to run AI inference workloads on dedicated NVIDIA GB200 NVL72 clusters via the CoreWeave Cloud platform. This deal includes deploying Perplexity Enterprise Max and running workloads on CoreWeave Kubernetes Service with W&B Models for model training and management.
1. Multi-Year Partnership with Perplexity
CoreWeave has entered a multi-year agreement with Perplexity to manage high-volume AI inference workloads on its Cloud platform, marking a significant enterprise contract and revenue stream for the company.
2. Dedicated NVIDIA GB200 NVL72 Infrastructure
Under the agreement, Perplexity will run inference operations on dedicated clusters powered by NVIDIA GB200 NVL72 GPUs, designed to handle the compute demands of its Sonar and Search API services at scale.
3. Platform Deployment and Services
Initial deployment sees Perplexity running workloads on CoreWeave Kubernetes Service with W&B Models, enabling end-to-end training, fine-tuning and production management, while deploying Perplexity Enterprise Max for internal search and advanced analysis.
4. Strategic Positioning and Industry Recognition
The partnership underscores CoreWeave’s leading AI cloud performance, building on top MLPerf and SemiAnalysis ClusterMAX benchmarks and aligning with NVIDIA’s initiative to establish over 5GW of AI factories by 2030.