Graph database platform for connected data and relationships
Neo4j is a Swedish-founded graph database platform that stores and queries data as nodes and relationships rather than rows and columns. The industry leader in graph technology, Neo4j powers fraud detection, recommendation engines, knowledge graphs, and network analysis at enterprises worldwide. Founded in Malmö in 2007, the company now operates under Neo4j, Inc. (San Mateo, CA) with Neo4j Sweden AB maintaining a major engineering presence in Malmö.
Headquarters
Malmö, Sweden
Founded
2007
Pricing
EU Data Hosting
Yes
Employees
501-1000
Open Source
Yes
Free
Pay-as-you-go
Pay-as-you-go
Contact Sales
Billing: pay-as-you-go, annual
Before Neo4j existed, developers trying to model relationships — social connections, fraud rings, supply chains — had to torture relational databases into doing something they were never designed for. In 2007, three engineers in Malmö, Sweden decided that was absurd. Emil Eifrem, Johan Svensson, and Peter Neubauer built Neo4j as a purpose-built graph database, storing data as nodes and relationships rather than rows and columns.
Nearly two decades later, Neo4j is the undisputed market leader in graph databases. It processes billions of relationships for companies like Volvo, Comcast, NASA, and UBS. The company has raised over $580 million, surpassed $200 million in annual recurring revenue, and employs roughly 1,000 people across offices in Malmö, San Mateo, London, and Munich.
There is a caveat for European buyers, though. While Neo4j was born in Sweden and Neo4j Sweden AB still operates a major engineering hub in Malmö, the corporate parent — Neo4j, Inc. — is incorporated in California. That matters for procurement teams evaluating data sovereignty. The technology roots are Swedish; the corporate structure is American. AuraDB's EU region hosting and ISO 27001/SOC 2 certifications close much of that gap, but it is worth understanding before signing a contract.
Cypher is Neo4j's declarative query language, and it remains the single best reason to choose this platform. Where SQL forces you to think in tables and joins, Cypher lets you express graph patterns visually: (person)-[:KNOWS]->(friend). The learning curve is gentler than Gremlin (Amazon Neptune) or SPARQL, and the language is now an open standard (GQL/ISO) adopted by other databases. For teams without graph database experience, Cypher dramatically reduces time-to-productivity.
AuraDB is Neo4j's fully managed cloud offering, available on AWS, GCP, and Azure across 60+ regions — including Frankfurt, Dublin, Amsterdam, and other EU locations. The free tier supports up to 200,000 nodes and 400,000 relationships, which is enough for prototyping but not production. Professional instances start at $65/GB/month with hourly metering, meaning a 4GB instance costs roughly $260/month. Business Critical adds a 99.95% SLA, automatic failover, and 24/7 support at $146/GB/month.
The pricing model is clean — you pay for provisioned memory, and compute, storage, backups, and network are included. No hidden I/O charges. That said, graph databases are memory-hungry by nature. A dataset that fits in 2GB of PostgreSQL might need 8-16GB of Neo4j RAM due to the index structures. Budget accordingly.
Neo4j's Graph Data Science (GDS) library includes over 130 algorithms — PageRank, community detection, node similarity, pathfinding, centrality, and link prediction. These run natively inside the database, avoiding the latency and complexity of exporting data to external analytics tools. For fraud detection, recommendation engines, and network analysis, GDS eliminates an entire layer of ETL.
Neo4j has leaned hard into the AI knowledge graph space. Native vector indexes support retrieval-augmented generation (RAG) workflows, allowing you to store embeddings directly alongside graph structures. The MCP (Model Context Protocol) server lets AI agents interact with Neo4j through natural language. LangChain integration is first-class. If you are building an AI application that needs structured knowledge alongside unstructured retrieval, Neo4j is one of the few databases that handles both natively.
The Enterprise Edition adds clustering (causal clusters with read replicas), role-based access control, LDAP/Active Directory integration, online differential backups, and Cypher query performance roughly 50-100% faster than Community Edition. These are table-stakes features for production deployments, but they are locked behind a commercial licence — a common friction point.
Neo4j's pricing has two distinct tracks. The self-hosted Community Edition is free under GPLv3, but lacks clustering, RBAC, and online backups. Self-hosted Enterprise requires a negotiated licence — expect annual contracts starting in the low five figures for small deployments.
AuraDB cloud is where most new customers land. The free tier is useful for learning but impractical for real work (200k node cap, paused after 30 days of inactivity). Professional starts at $65/GB/month — a 4GB instance runs roughly $3,120/year. Business Critical at $146/GB/month doubles that, but adds the SLA and support that enterprise buyers require.
Compared to Amazon Neptune (which charges separately for I/O, storage, and compute), Neo4j's all-inclusive GB pricing is easier to predict. Compared to PostgreSQL with a graph extension, Neo4j is significantly more expensive — but it is also significantly more capable for graph-native workloads. The value equation depends entirely on whether your data is genuinely relationship-heavy.
Neo4j holds SOC 2 Type II, ISO 27001:2022, and HIPAA certifications. AuraDB instances can be deployed in EU regions on all three major cloud providers, and data residency is guaranteed within the selected region. Encryption is enforced at rest and in transit.
The structural complication is corporate. Neo4j Sweden AB operates from Malmö and contracts under Swedish law for European customers. But the parent entity, Neo4j, Inc., is a Delaware/California corporation. Under CLOUD Act, US authorities could theoretically compel data disclosure from a US parent — even for EU-hosted data. Neo4j mitigates this with EU-region isolation and contractual commitments, but organisations in highly regulated sectors (banking, defence, healthcare) should evaluate this carefully with legal counsel.
For most European businesses, Neo4j's compliance posture is strong. The certifications are current, the EU hosting is genuine, and the Swedish engineering presence provides a meaningful European footprint. It is not the same as a fully EU-owned company, but it is substantially better than most US-headquartered database vendors.
Data teams modelling complex relationships — fraud rings, supply chains, identity graphs, social networks. If your queries routinely involve three or more hops across relationships, a graph database is not optional; it is necessary. Neo4j is the safest choice in this space.
AI and knowledge graph builders using RAG, vector search, or LLM-powered applications that need structured knowledge. Neo4j's native vector indexes and LangChain integration make it a natural fit.
Enterprises needing EU-hosted graph infrastructure with certifications that satisfy compliance teams. AuraDB's EU regions and Neo4j Sweden AB's contractual structure cover most requirements.
Not ideal for: teams that need a general-purpose database, simple key-value storage, or are budget-constrained. Neo4j is a specialist tool with specialist pricing.
Neo4j remains the gold standard for graph databases, and nothing else in the market matches its combination of query language maturity, ecosystem depth, and enterprise readiness. The Swedish engineering heritage is real, and AuraDB's EU hosting is genuine. The ownership structure — a US parent with a Swedish subsidiary — is a legitimate consideration, not a dealbreaker. For European organisations working with relationship-heavy data, Neo4j is the obvious first choice. Just make sure your workload actually needs a graph database before committing to the pricing.
Yes. Neo4j holds SOC 2 Type II and ISO 27001:2022 certifications. AuraDB offers EU-region hosting on AWS, GCP, and Azure with data residency guarantees. European customers contract through Neo4j Sweden AB under Swedish law. The parent company (Neo4j, Inc.) is US-based, which introduces CLOUD Act considerations that highly regulated organisations should assess.
The Community Edition is free and open-source (GPLv3), suitable for single-instance deployments with full ACID transaction support. AuraDB Free provides a cloud instance with up to 200,000 nodes and 400,000 relationships at no cost. Neither is suitable for production workloads requiring clustering or enterprise support.
Neo4j offers Cypher (more intuitive than Gremlin/SPARQL), a larger third-party ecosystem, self-hosting options, and a mature Graph Data Science library. Neptune integrates more tightly with AWS services and is fully managed without self-hosting overhead. On pure graph traversal performance, Neo4j generally leads in read-heavy workloads, while Neptune can be simpler to operate within an all-AWS architecture.
Partially. Neo4j was founded in Malmö, Sweden in 2007, and Neo4j Sweden AB (179 employees, Malmö) remains the entity that contracts with European customers. However, the parent company Neo4j, Inc. is incorporated in the United States with headquarters in San Mateo, California. The technology and a significant portion of the engineering team are Swedish; the corporate governance is American.
Neo4j is a specialist graph database, not a general-purpose data store. It performs poorly on simple CRUD operations, tabular analytics, time-series data, or workloads with few relationships between records. If your data fits naturally into rows and columns, PostgreSQL or MySQL will be cheaper, faster, and simpler. Use Neo4j when relationships are the query, not just the data.
Managed open-source data infrastructure in the cloud
High-performance analytics database built for speed and in-memory processing
Alternative to Amazon Redshift, Oracle, Snowflake