Qdrant Raises $50M Series B to Scale Rust-Built Vector Search Infrastructure

Qdrant, the open-source vector search engine written entirely in Rust, closed a $50 million Series B on March 12, led by AVP (Advance Venture Partners) with participation from Bosch Ventures, Unusual Ventures, Spark Capital, and 42CAP. The round brings total capital raised to $87.8 million, roughly doubling the $28 million Series A Unusual Ventures led in early 2024.

The Berlin and New York City-based company has staked its entire architecture on a decision made before its first line of production code: build in Rust, control the stack down to the assembly level, and expose every layer of retrieval as a composable primitive. That means engineers working with Qdrant aren't accepting opaque defaults. They combine dense vectors, sparse vectors, metadata filters, multi-vector representations, and custom scoring functions at query time, with explicit dials for ranking, indexing, latency, and cost trade-offs. The official positioning is blunt: "rethink every layer of the retrieval — indexing, scoring, filtering, ranking — as composable primitives that engineers control directly."

Co-founder and CEO Andre Zayarni framed the production case plainly: "Retrieval runs within agent loops, executing thousands of queries per workflow against continuously changing data. We built Qdrant to create foundational infrastructure for the AI era, giving developers the controls they need to optimise performance, latency, and cost without re-architecting the system."

That framing matters for the Rust community specifically. Unusual Ventures, which has backed Qdrant since the Series A, described the architectural bet as: "build from the ground up in Rust, control the stack down to assembly, and treat every aspect of retrieval as a composable primitive that engineers control directly. The result is a vector search engine that adapts to the problem rather than forcing the problem to fit the tool." The low-level memory control and predictable performance characteristics that make Rust attractive for systems programming translate directly into what Qdrant claims is predictable, low-tail latency at billion-scale across cloud, hybrid, on-premises, and edge deployments.

Deployment is flexible by design. Developers can spin up the binary locally, run it in a Docker container for quick testing, self-host on Kubernetes, use the fully managed Qdrant Cloud, or go hybrid or private cloud. The engine is fully open-source, which according to Unusual Ventures has generated 250 million-plus downloads and 29,000 GitHub stars. The investor blog also names Canva, HubSpot, Bosch, Tripadvisor, and OpenTable as businesses running Qdrant in production.

AVP's Warda Shaheen positioned the investment in infrastructure-cycle terms: "With every infrastructure shift...purpose-built systems emerge and rapidly scale in fast-growing new markets...[W]e're seeing this pattern again with Qdrant...at the forefront of building the retrieval layer of the future."

The largest share of the $50 million is earmarked for engineering, with Zayarni citing two focal points: improving retrieval for agentic workloads and helping enterprises control AI costs as those workloads scale. Go-to-market efforts will focus on growing the Qdrant Cloud user base, and hiring will target both engineers and customer-facing roles.

Analyst Catanzano, quoted by TechTarget, noted that closing a round of this size in the current climate carries its own signal: "Qdrant's ability to secure funding in a cautious VC climate highlights its strong value proposition and market relevance. It demonstrates investor confidence in their Rust-based architecture and composable vector search approach, which addresses critical needs in AI-driven retrieval infrastructure."

Qdrant is also listed in the Forrester Wave for Vector Databases, according to Unusual Ventures. For engineers already sold on Rust's performance and safety guarantees, watching those properties carry a vector database to $87.8 million in total funding and production deployments at Canva scale is a pretty reasonable validation of the language choice.