CoreWeave has inked a multi-year deal with Perplexity to host inference workloads for the AI service provider.
The partnership establishes the cloud platform as the primary infrastructure provider for Perplexity’s AI suite. A CoreWeave press release announcing the deal did not disclose the exact length or value of the deal.
Perplexity will power its inference workloads using NVIDIA GB200 NVL72-powered clusters, focusing on the growth of its Sonar and Search API ecosystem.
Per the agreement, CoreWeave will integrate Perplexity Enterprise Max across its organization. This enterprise software rollout gives employees access to Perplexity’s advanced AI models for research and data analysis tools.
The partnership is the latest feather in the cap of Nvidia’s growing neocloud business.
Nvidia invested $2 billion in CoreWeave on January 26 to accelerate data center expansion. The equity stake funds the procurement of land, power, and infrastructure for new AI data centers.
The company faces ongoing operational pressures to bring capacity online for massive AI workloads. CoreWeave lowered its 2025 capital expenditure guidance to $12 billion to $14 billion on December 16, 2025, and severe summer weather caused construction delays at Core Scientific’s 260-megawatt site in Denton, Texas, which CoreWeave is contracted to use for one of its GPU deployments.
CoreWeave attempted to acquire Core Scientific for $9 billion on October 30, 2025. The all-stock merger failed after Core Scientific shareholders rejected the deal.
Shares of CoreWeave (NASDAQ: CRWV) traded up 7.8% to $79.50 in morning trading hours on the news.
