SurrealDB vs EdgeDB: Comparing multi-model and relational engines

The distinction between relational and multi-model engines

SurrealDB and EdgeDB occupy different spaces in the 2026 database market. SurrealDB uses Rust to combine document, graph, relational, time-series, geospatial, and key-value models. EdgeDB relies on Postgres to provide relational capabilities with foreign key support and light graph functions. I find the claim that EdgeDB acts as a true graph database to be a stretch because it lacks recursion support. It manages connections through the application layer rather than the system layer, which creates performance bottlenecks when users hit maximum link depths. Does EdgeDB plan to solve these recursion limitations before its next major version? You should understand these architectural differences before selecting a stack for your next project.

EdgeDB is an awesome ORM driven relational database for setups that require light graph capabilities. However, the claim that a Postgres wrapped EdgeDB is a graph database is more an investor pitch than a technical foundation. While you can mimic graphs with a relational database, it is not built for that. It is not a classic ORM because it does not perform query result post-processing at the application level. Every EdgeQL query compiles into a single SQL query, which returns the results to EdgeDB clients directly. EdgeDB landed $15 million in November before launching its cloud database product.

Scaling through multi-model integration

SurrealDB recently secured a $23 million extension to its Series A round, led by Madrona. This brings its total funding to approximately $44 million. The company targets AI agents and real-time applications by unifying diverse data types. Version 3.0 provides upgraded vector search and indexing for unstructured data workloads. The architecture separates compute and storage to allow horizontal scaling. This design helps the platform scale to hundreds of nodes for high availability and capacity. The company was bootstrapped for three years before closing a $6 million seed round led by FirstMark.

SurrealDB integrates document, graph, relational, time-series, geospatial, and key-value models into a single engine that supports full-text, vector, hybrid, real-time, and event-driven capabilities for various complex and diverse application workloads. The company targets the growing market for database-as-a-service offerings, which may reach $24.8 billion by 2025. SurrealDB competes with vendors like MongoDB, Neo4j, Couchbase, and DynamoDB, as well as back-end service providers like Firebase, Supabase, and Nhost. The three-person team aims to increase headcount to 18 within the next year. SurrealDB also handles business logic and user authentication directly within the database layer. This reduces the need for custom backend code. Users can connect records with directed graph edge connections without using JOINs. This approach eliminates the N+1 query problem.

Developer experience and build constraints

Developers using SurrealDB in Rust projects report significant build times. One user noted that a full workspace build on a MacBook M3 took over five minutes. Half of that time went to building SurrealDB crates. A build on a DigitalOcean droplet lasted over 50 minutes and crashed due to memory shortages. On an Intel core i7-12700K, debug builds took 27 seconds and release builds took 37 seconds. The crate contains approximately 1.5 million lines of code.

SurrealDB offers several deployment options. You can run it as a single in-memory node or as part of a distributed cluster. It works as an embedded library or a standalone server. The system uses specific techniques for handling multi-table transactions and document record IDs. These techniques allow it to operate without table or row locks. Users can deploy it via Homebrew, Docker, or Kubernetes. The platform provides native SDKs for Python, JavaScript, Rust, Golang, Java, .NET, PHP, C, Dart, and Swift. It also supports WebAssembly for use in the browser.

EdgeDB developers focus on relational setups. The platform allows for strongly-typed data through its schema model. It compiles every EdgeQL query into a single SQL query for the Postgres engine. This ensures that queries return results directly to clients without extra post-processing. The company landed $15 million in November before launching its cloud product.