Skip to main content

Kuzu

Category: AI in Analytics & BI

An open-source, embeddable property graph database management system (GDBMS) designed for high-speed analytical workloads and seamless integration into the modern data science stack. Kuzu was founded in 2023. The company is led by Semih Salihoglu. Based in Waterloo, Canada. Team size: 11. Total funding raised: Undisclosed. Latest round: Acquisition. Key investors include ["Felicis","Unusual Ventures","Essence Venture Capital","The Accelerator Centre","Apple"].

Founded
2023
Headquarters
Waterloo, Canada
Team size
11
Total funding
Undisclosed

Value proposition

Provides 'DuckDB-like' ease of use for graph data by being an in-process database that eliminates server management while delivering massive performance gains for complex many-to-many joins through advanced query optimization.

Products and solutions

["KùzuDB Core Engine (C++ based, embeddable)","Kùzu Explorer (Web-based graph visualization and schema management tool)","Built-in Vector Search & Full-Text Search indices","Multi-language APIs (Python, Rust, C++, Node.js, Go, Java)","GraphRAG Integration Suite (LangChain, LlamaIndex, and Haystack connectors)"]

Unique value

It is the first graph database to successfully implement an 'in-process' architecture that combines columnar storage with 'Worst-Case Optimal Join' (WCOJ) algorithms, making it uniquely suited for local, high-performance graph analytics.

Target customer

AI engineers, data scientists, and software developers building GraphRAG (Retrieval-Augmented Generation) applications, recommendation engines, and fraud detection systems.

Industries served

["Artificial Intelligence & Machine Learning","Financial Services (Fraud Detection & Anti-Money Laundering)","Cybersecurity (Threat Intelligence & Network Analysis)","Healthcare & Life Sciences (Knowledge Graphs & Drug Discovery)","E-commerce (Personalized Recommendation Systems)"]

Technology advantage

Utilizes 'Factorized Query Execution' to minimize data redundancy during processing and WCOJ to outperform traditional binary joins on cyclic queries by orders of magnitude. Its zero-copy integration with the PyData ecosystem (Pandas, Polars, Arrow) allows for instant data transfer between the database and analytical tools.

How they differentiate

Kuzu differentiates by being an 'in-process' (embeddable) database that requires no server management, utilizing columnar storage and 'Worst-Case Optimal Join' (WCOJ) algorithms to outperform traditional graph databases on complex, cyclic queries.

Main competitors

["Neo4j","Memgraph","DuckDB"]

Key partnerships

["Apple (Parent Company/Acquirer)","University of Waterloo (Research & Talent Pipeline)","PyData Ecosystem (Integration with Arrow, Pandas, and Polars)","AI Framework Providers (LangChain, LlamaIndex)"]

Notable customers

["LangChain (Integration)","LlamaIndex (Integration)","Haystack (Integration)"]

Major milestones

["Commercial spinoff from the University of Waterloo Data Systems Group in 2023","Secured Seed funding led by Felicis in 2023","Released Kùzu Explorer for graph visualization in 2024","Acquired by Apple in October 2025 to bolster on-device AI and data infrastructure"]

Growth metrics

Estimated revenue of $1.2M in 2025 with a lean team of 11 employees prior to acquisition.

Market positioning

High-performance embeddable graph OLAP for AI engineers and data scientists.

Geographic focus

North America (HQ in Canada), Global Open Source community

Patents and IP

Proprietary technology and IP derived from seminal research conducted at the University of Waterloo's Data Systems Group; specific patent filings are typically held under the university or the acquired entity.

About Semih Salihoglu

Semih Salihoglu is an Associate Professor of Computer Science at the University of Waterloo and a leading expert in graph database systems. He holds a PhD from Stanford University, where he contributed to the Pregel graph processing system. Before co-founding Kuzu to commercialize a high-performance, embeddable graph database, he conducted extensive research on worst-case optimal joins and factorized databases. His leadership led to the company's acquisition by Apple in late 2025.

Official website: