Highlights:

  • Zilliz is primarily responsible for creating the open-source Milvus vector database that is used to back up AI applications.

Zilliz Inc, the startup best known for developing the Milvus vector database, revealed that it had raised USD 60 million in an extension of its previous Series B fundraising round of USD 43 million.

The current round was led by Prosperity7 Ventures, the diversified growth fund of Aramco Ventures. Existing investors such as Temasek’s Pavilion Capital, Hillhouse Capital, 5Y Capital, and Yunqi Capital participated as well. To date, the firm has raised USD 113 million in funding.

Zilliz is primarily responsible for developing the open-source Milvus vector database that is used to back up Artificial Intelligence (AI) applications. A vector database design is considered superior for AI models, especially machine learning.

These models typically transform unstructured data like documents, movies, and user actions into vectors, which are essentially complex series of numbers. Therefore, determining which vectors are most similar is typically a necessary step in making inferences.

A specialized vector database is necessary to sort and rank a large number of vectors. Conventional databases aren’t well-suited to machine learning because they focus on tables and documents.  Milvus can index and transform millions of vectors dynamic and is, thus, well-suited to answering the questions typically asked of AI models.

Zilliz founder and Chief Executive Charles Xie stated in an interview, “Modern AI algorithms use high-dimensional vectors called embeddings to represent the deep semantics of unstructured data — necessitating a new form of data infrastructure to manage and process them at scale. He added, “Vector databases fulfil this purpose, providing a unified way to store, index and search across massive quantities of high-dimensional vectors, and further offloading the burden of complex data management from AI application developers.”

Milvus is open source but setting it up and keeping it functioning smoothly calls for a high level of knowledge. Ziliz fills this role perfectly. Zilliz Cloud, which the firm calls “a high-performance vector database management system,” is a managed version of Milvus. Zilliz emphasizes that the solution can handle billions of vectors in a matter of milliseconds and is native to the cloud.

Xie added, “Zilliz Cloud is a fully managed database-as-a-service built on the open-source vector database Milvus we created at Zilliz. Using Milvus as the underlying core, Zilliz Cloud provides an integrated platform for vector data processing, unstructured data analytics and enterprise AI application development.”

Andy Thurai, vice president and principal analyst at Constellation Research Inc., said that Zilliz’s open-source approach sets it apart from the vector database pack as most competing products are built on underlying proprietary platforms. Additionally, Zilliz is more expandable than competing products. That, according to Thurai, is a crucial capacity because it’s not enough just to store vectorized unstructured data. The information must also be searchable.

He explained, “AI and machine learning models need to be able to search unstructured content in real-time. Enterprises have struggled with searching unstructured data such as images, audio files and PDFs for a long time, and AI has the potential to solve this issue. So, the data of an image can be vectorized as an array of numbers or an AI-generated representation of that image. Then, a vector search can be performed on a vector database that stores such information and returns either the top search results of the match or all results above a certain threshold.”

According to Thurai, Zilliz’s capability to do searches on unstructured data leads to a wide variety of use cases, including but not limited to searching for a specific image, facial recognition, and voice string matching. He said that businesses now have a mechanism to look for specific information in the mountains of unstructured data they have been gathering and holding for years without ever using it.

Thurai said, “Given the importance of unstructured data search, we’re seeing a lot of investments in small startups offering these kinds of capabilities. It is an uphill battle with all the competition, but open source always wins with traction. Proper monetization of these technologies can be hit or miss, though. Hopefully, Zilliz will gain traction and find a way to monetize its solution soon.”

Private preview access to Zilliz’s paid service is presently accessible by invitation only so that a small number of clients can test and provide feedback before the product’s full public release early next year. Zilliz has said its long-term goal is to have Zilliz Cloud mature into a fully managed Database-as-a-Service offering that serves as a unified platform for vector data processing, unstructured data analytics, and enterprise AI application development.

Xie added, “Milvus has now become the world’s most popular open-source vector database with over a thousand end-users. We will continue to serve as a primary contributor and committer to Milvus and deliver on our promise to provide a fully managed vector database service on the public cloud with the security, reliability, ease of use, and affordability that enterprises require.”

Zilliz plans to use the capital from today’s round to grow its engineering and go-to-market teams to further its commitment to open source and develop its managed cloud solution.