Milvus (Zilliz)

Zilliz runs a freemium model built on the open-source Milvus vector database, converting community users into paying cloud and enterprise customers. The open-source project — an LF AI Foundation graduate — is free to download and self-host, which drives adoption across AI teams building recommendation engines, computer vision pipelines, and retrieval-augmented generation applications. Revenue comes from two paid paths: Zilliz Cloud, a fully managed SaaS offering where customers are billed on compute and storage consumption, and enterprise licenses that add security, dedicated support, and on-premises deployment options for large organizations. The open-source layer acts as a demand-generation engine and a moat — engineers evaluate Milvus for free on their own infrastructure, and a subset graduates to the managed service when they need operations at scale. The cloud side carries infrastructure-like unit economics typical of consumption-based SaaS: gross margins are driven by compute and storage density, though Zilliz must pay for underlying cloud infrastructure. The enterprise license tier mirrors standard B2B SaaS with annual contracts and per-node or cluster-based pricing. No public revenue data is available, but the model structurally benefits from the rising volume of vector search workloads as AI applications enter production.