Zilliz, a vector database company, harvests USD 60 million (CNY 411 million) in a Series B+ financing round.
Vector database company Zilliz announced that it closed a US$60 million Series B+ round of funding on August 25. The funding was led by Prosperity7 Ventures, a subsidiary of Aramco Ventures, and followed by existing shareholders Pavilion Capital, Hillhouse, 5Y Capital and Yunqi Partners, with InvesTarget acting as an exclusive financial advisor, news portal lieyun.com reported.
Aysar Tayeb, Executive Managing Director of Prosperity7 Ventures, said: "Unstructured data is already more than four times the total amount of structured data, and as technologies including human intelligence, robotics and the Internet of Things converge in the real and digital worlds, the share of unstructured data in the global data universe will further expand." This explains why Zilliz gained its support.
Zilliz, founded in 2017 focuses on developing vector databases for AI scenarios. Traditional databases including Oracle and PostgreSQL mainly deal with structured data, which is relatively fixed in size and format and has limited data dimensions. While AI models usually deal with unstructured data including images, videos, speech and text which should be transformed into vectors before being input. These vectors range from a few hundred to tens of thousands of dimensions. Traditional database techniques such as storage and indexing (index) are not suitable for such high-dimensional data. Therefore, there is a need for a dedicated vector database product for the AI domain, Zilliz believes.
Milvus, the vector database developed by Zilliz, is more closely aligned with the needs of AI production systems. It has been designed with a vector data architecture that allows it to run nearly a thousand times faster and scale up to milliseconds compared to the traditional PostgreSQL database, which is capable of processing billion-sized vector data. It has also designed new query algorithms for different CPU and GPU platforms to speed up the computation of multidimensional data.
Zilliz’s international competitors include Tableau and Microsoft and domestic ones include Yusur (Chinese:中科驭数), Novumind(Chinese:异构智能) and Createlink Technology (Chinese:创邻科技).