I previously wrote about the importance of knowledge graphs to data intelligence and artificial intelligence (AI). A knowledge graph can surface information about the relationships and dependencies between virtual and physical objects by providing a structured representation of data that identifies the connections between entities and attributes. This representation of enterprise knowledge can be used to enhance enterprise understanding of data and provide an additional layer of information to accelerate operational, analytic and AI initiatives. As interest in knowledge graphs grows, graph database specialist Neo4j is adapting its product portfolio to facilitate the creation of knowledge graphs and highlight the potential benefits of natively storing data as entities, attributes and relationships.
Neo4j was founded in 2007 to build a business around the open-source graph database of the same name. The company’s Neo4j Graph Database enables data to be stored using a 
property graph model, which natively stores information about entities (as nodes), the relationships between them (as edges) and their values (as properties). As I previously explained, although the graph data model is less widely adopted than the relational model, which has dominated the data platform sector for the past 50 years, graph databases are inherently more suitable than relational databases for large-scale and high-performance navigation of relationships between entities. Neo4j enjoyed early success for use cases including network management, master data management, social media, fraud detection and navigation systems and now claims more than 1,700 customers, including 84 of the Fortune 100. The company announced in November 2024 that it had surpassed $200 million in annual recurring revenue and secured a $50 million investment from Noteus Partners, adding to the $325 million Series F funding round it announced in June 2021, which valued the company at over $2 billion. The Neo4j Graph Database is available for self-managed deployment on-premises, in the cloud or via the Neo4j AuraDB managed cloud service. Additionally, the company offers capabilities for graph analytics, data science and visualization. Enterprise interest in knowledge graphs is accelerating, and Neo4j is keen to highlight the potential benefits of a native graph data model and position its combined capabilities as a Graph Intelligence Platform that can be used to accelerate the development of AI and agentic AI applications.
Despite its name, the relational data model is not particularly efficient when it comes to identifying relationships between entities and attributes. With data stored in tables of rows and columns, a combination of normalization, row and column keys and table joins is required to identify relationships. These are common and widely adopted approaches, but they come with a performance impact, especially at scale. In comparison, graph databases enable more efficient identification and navigation of connections between data by natively storing relationships between entities as well as their attributes. This makes the graph model ideal for supporting use cases such as navigation systems, social media and fraud detection as well as content and knowledge management based on semantic models. Adoption of graph databases remains relatively niche, especially compared to the relational model, but proponents such as Neo4j argue that potential use cases are far more prevalent than most enterprises realize. The company has identified seven graphs of the enterprise that are dependent on relationships: employees, suppliers, finance, customers, processes, products, and networks and security. The argument is that every enterprise is built and dependent upon graphs, but most don’t model their data accordingly.
As well as targeting individual graph use cases, Neo4j is also looking to exploit growing demand for enterprise-wide knowledge graphs to provide an abstraction layer that surfaces
the relationships between distributed and disparate data, boosting AI projects by enabling the identification of relationships and dependencies between entities and attributes. Graph databases are not a prerequisite for creating a knowledge graph. Nevertheless, Neo4j maintains that native graph storage and processing can accelerate the creation and performance of knowledge graphs for AI and recently announced an investment of $100 million over three years to boost its capabilities as a platform for agentic AI applications. Some of that investment will be allocated to sales and marketing and ecosystem integration, as well as a startup program to attract new customers and partners, but the majority is being earmarked for product innovation. A key focus is on facilitating the development of agentic applications. Early deliverables include Neo4j Aura Agent, which is designed to enable users to build, test and deploy AI agents and is now in early access, as well as the launch of a Model Context Protocol (MCP) Server for Neo4j, which is designed to integrate graph-based memory and reasoning into agent or AI applications. MCP has recently emerged as a key open standard for enabling agentic AI by providing connectivity between agents and the data needed to support automated action execution. I assert that through 2027, almost all data-related software providers will adopt MCP to provide interoperability between agentic applications and trusted enterprise data and business workflows. Neo4j also plans to enhance the Neo4j Graph Database to boost support for storing and indexing vectors to facilitate GraphRAG use cases that augment foundation models with information based on logical connections between entities and attributes in addition to semantic similarity. The company has already taken steps to enhance the scalability of its core database engine with the launch of its Infinigraph distributed graph architecture, which is designed to deliver horizontal scaling to support workloads of more than 100TB. Infinigraph builds on the company’s existing autonomous clustering approach and introduces a new property sharding capability that separates a single graph shard containing the topology, nodes, relationships, labels and unique identifiers from multiple distributed property shards that store the properties of the data.
Neo4j has taken substantial steps to enhance its portfolio for graph analytics and AI in recent years, and its commitment to facilitate knowledge graphs through additional investment in the development of agentic applications is significant, along with the recent addition of the Infinigraph architecture. The investment commitment is just the first step, and it remains to be seen how the money will be spent, but in the meantime, I recommend that enterprises evaluating potential use cases for the graph data model and knowledge graphs include Neo4j in their assessments.
Regards,
Matt Aslett
Fill out the form to continue reading.