Milvus was open-sourced in 2019 — three full years before 'vector database' became a buzzword — and it was already a Linux Foundation top-level project by 2021. The company behind it, Zilliz, raised over $70 million to build a cloud offering on top. When the LLM wave hit in 2023 and enterprise teams scrambled for production-grade vector search, Milvus was the most battle-tested option available. Being early and patient in infrastructure is rare. It paid off.
Founded
2019
HQ
San Francisco, California
Total Raised
$113 million (Zilliz)
Founder
Charles Xie, Xiaofan Luan
Status
Private (backed by Zilliz)
Website
milvus.ioTHE ORIGIN STORY
Charles Xie and Xiaofan Luan founded Zilliz in 2017 in San Francisco with a specific thesis: as AI and machine learning matured, the bottleneck would shift from model training to data retrieval. Finding the right data — fast, at scale — would become a core infrastructure problem that existing databases were not built to solve.
They built Milvus, an open-source vector database, to close that gap. Milvus was released publicly in October 2019 under the Apache 2.0 license.
Within two years it had become one of the most starred AI infrastructure projects on GitHub. In 2021, the Linux Foundation's LF AI & Data project accepted Milvus as a top-level project — putting it in the same tier as TensorFlow and PyTorch in the AI ecosystem.
Zilliz built Zilliz Cloud, a fully managed Milvus service, as the commercial layer on top.
WHAT THEY ACTUALLY DO
Milvus is free and open source. Zilliz monetizes it through Zilliz Cloud — a fully managed Milvus deployment hosted on AWS, GCP, and Azure.
Enterprise customers pay for managed infrastructure, SLA-backed support, and private cloud deployments. The open-source project drives global developer adoption.
Zilliz converts a fraction of those developers into enterprise contracts, where the deal sizes are large and the switching costs are high.
THE PRODUCTS
Milvus (open source) — the flagship vector database supporting billion-scale vector similarity search. Supports multiple index types including HNSW, IVF, and DiskANN.
Zilliz Cloud — the fully managed enterprise Milvus service. Attu — an open-source GUI for managing and visualizing Milvus collections.
PyMilvus — the Python SDK and the most popular way developers interact with the database.
HOW THEY GREW
Community first, enterprise second. Milvus invested years in the developer community before pushing commercial sales.
The team contributed to LF AI & Data, published rigorous technical documentation, and appeared consistently in academic ML papers as a reference implementation. When ChatGPT launched in late 2022 and enterprise teams scrambled to build retrieval pipelines, Milvus already had the credibility, the documentation, and the production track record.
The groundwork was years old before the market caught up.
THE HARD PART
Being the original does not automatically make you the market leader. Pinecone, Chroma, Qdrant, and Weaviate all gained significant developer mindshare during the 2023 AI boom.
Milvus has the deepest feature set and the longest production track record, but the developer market rewards simplicity and fast onboarding — and newer entrants are often easier to get started with. The challenge is winning developer love without compromising the enterprise capabilities that justify the complexity.
MONEY TRAIL
Series A
2020 · Led by Yunqi Capital
$13M raised
Series B
2022 · Led by Prosperity7 Ventures
$60M raised
WHO BACKED THEM
Yunqi Capital, Prosperity7 Ventures (Aramco subsidiary), Hillhouse Capital
Related Profiles
Companies
Databricks
Databricks and Milvus are complementary enterprise AI layers. Databricks handles data processing and model training. Milvus handles vector retrieval. Many enterprise ML teams use both in the same production stack.
Snowflake
Snowflake customers frequently layer Milvus for vector search on top of their Snowflake data. The two platforms represent storage for different data types — structured in Snowflake, unstructured embeddings in Milvus — within the same enterprise AI architecture.
Head-to-Head
Compare Milvus vs another company.