Quadrantthe company behind the open source vector database of the same name, has raised $28 million in a Series A funding round led by Spark Capital.
Founded in 2021, Berlin-based Qdrant seeks to capitalize on the rapid AI revolution by targeting developers with an open-source vector search engine and database — an integral part of genetic AI, which requires building relationships between unstructured data (eg text, images, or audio that is not tagged or otherwise organized), even when that data is “dynamic” in real-time applications. According Gartner dataunstructured data makes up about 90% of all new enterprise data and is growing three times faster than its structured counterpart.
The vector database field is hot. In recent months we have seen such Weaviate raised $50 million for its open source vector database, while Zilliz secured $60 million to commercialize its Milvus open source vector database. Other Color secured $18 million in seed funding for a similar proposal, while Pinecone raised $100 million for a proprietary alternative.
Qdrant, for its part, raised $7.5 million last April, further underscoring the seemingly insatiable appetite investors have for vector databases — while also signaling a planned expansion on Qdrant’s part.
“The plan was to go ahead with the next fundraising in the second quarter of this year, but we received an offer a few months earlier and decided to save time and start scaling the company now,” explained Qdrant CEO and co-founder Andre Zayarni . TechCrunch. “Raising money and hiring the right people always takes time.”
Notably, Zayarni says the company actually turned down a potential takeover bid from a “major player in the database market” at the same time it received an investment offer. “We went with the investment,” he said, adding that they will use the new cash injection to build out its business team, given that the company is essentially made up of engineers right now.
Binary logic
In the nine months since its last raise, Qdrant has been successful launched a new highly efficient compression technology called binary quantization (BQ), which focuses on low-latency, high-performance indexing, which it says can reduce memory consumption by up to 32x and improve retrieval speeds by around 40x.
“Binary quantization is a way to ‘compress’ vectors into the simplest possible representation with only zeros and ones,” Zayarni said. “Comparing vectors becomes the simplest CPU command — this makes it possible to significantly speed up queries and dramatically save on memory usage. The theoretical idea is not new, but we have implemented it in such a way that there is very little loss of accuracy.”
However, BQ may not work for all AI models, and it’s entirely up to the user to decide which compression option works best for their use cases — but Zayarni says the best results they found were with his models OpenAI, while Cohere also worked as did Google’s Gemini. The company is currently being compared to models such as Mistral and Stability AI.
It is such efforts that have helped attract high-profile names including Deloitte, Accenture and — arguably the highest profile of all — X (genus Twitter). Or perhaps more accurately, Elon Musk’s xAI, a company developing ChatGPT competitor Grok, which debuted on the X platform last month.
Although Zayarni did not disclose details of how X or xAI used Qdrant due to a non-disclosure agreement (NDA), it is reasonable to assume that they use Qdrant to process real-time data. Indeed, Grok uses a generative AI model called Grok-1, trained on data from the web and human feedback, and given its (now) tight alignment with X, can incorporate real-time data from posts on social media in his responses. is what is known today as augmented recovery generation (RAG), and Elon Musk has publicly teased such use cases in recent months.
Qdrant doesn’t disclose which of its customers use its open-source incarnation of Qdrant and which use its managed services, but it pointed to a number of startups, including GitBook, VoiceFlow and Dust, which “mostly” use it . managed cloud service — this essentially saves resource-constrained companies from having to manage and develop everything themselves as they should with the open source core incarnation.
But Zayarni is adamant that the company’s open source credentials are one of its biggest selling points, even if a company chooses to pay for additional services.
“When you use a proprietary or cloud-only solution, there’s always the risk of vendor lock-in,” Zayarni said. “If the vendor decides to adjust pricing or change other terms, customers have to agree or consider switching to an alternative solution, which isn’t easy if it’s heavy production use. With open source, there’s always more control over your data and can switch between different deployment options.”
Alongside the funding today, Qdrant is also officially launching its managed ‘on-premise’ version, giving businesses the ability to host everything in-house but leverage the premium features and support that Qdrant provides. This follows last week’s news that the cloud version of Qdrand was landing on Microsoft Azure, adding to the existing AWS and Google Cloud Platform support.
In addition to lead backer Spark Capitali, Qdrant’s Series A round included participation from Unusual Ventures and 42 participations.
