The PostgreSQL Moment:How Open Source Is Rewriting the Database Power Structure

1. The Gap Is Almost Gone

For most of this decade, the question “will PostgreSQL ever crack the top three?” felt theoretical. The DB-Engines ranking — which aggregates job postings, search interest, and technical forum activity across hundreds of database systems — had kept Oracle, MySQL, and SQL Server locked in a seemingly immovable triad since the index’s inception in 2012.

June 2026 changed the tone of that conversation. ✓ Verified According to DB-Engines data confirmed by multiple independent trackers, SQL Server holds third place with approximately 698 points while PostgreSQL sits at fourth with roughly 688 — a gap of around 10 points. Twelve months ago, that gap was over 86 points.

What makes this trajectory remarkable is not simply the gap narrowing but the direction of every arrow. Among the top four, only PostgreSQL is growing. Oracle, MySQL, and SQL Server are all in decline. At the current pace of convergence — PostgreSQL gaining roughly 5–6 points a month while SQL Server sheds roughly 3 — a crossing into third place is projected by late Q3 or Q4 2026.

“The iron triangle of Oracle, MySQL, and SQL Server has held for two decades. PostgreSQL is on the verge of breaking it.”

If that happens, it will be the first structural change at the top of the global database rankings in over twenty years — a milestone no previous open-source project has achieved.

2. Why PostgreSQL? Three Structural Advantages

PostgreSQL’s rise is not a fluke of timing. It rests on three durable structural advantages that compound over time.

Open license, open ecosystem. PostgreSQL uses the permissive BSD license, which allows closed-source derivatives and commercial forks without restriction. MySQL’s GPL license carries “viral” terms that limit secondary commercial development. The licensing difference has led to a far richer PostgreSQL commercial ecosystem: every major cloud vendor now offers a managed PostgreSQL service (Amazon Aurora, Google AlloyDB, Azure Database for PostgreSQL, and now HorizonDB), while MySQL’s commercial future is increasingly owned by Oracle.

Cloud-native architecture alignment. PostgreSQL’s design — with its well-defined storage API, extensible type system, and clean separation of concerns — maps naturally onto storage-compute separation architectures that define modern cloud databases. As cloud adoption becomes the enterprise default, PostgreSQL becomes the path of least resistance.

pgvector: the AI killer app. The pgvector extension gives PostgreSQL native vector similarity search, enabling semantic search and retrieval-augmented generation (RAG) directly within the relational engine. This collapses a two-system architecture (relational database + separate vector store) into one, with all the operational simplicity that implies. Adoption has been explosive, with over 100 million downloads cited by ecosystem sources — a figure consistent with Supabase’s disclosed metric of managing over one million databases with 2,500 new ones spun up daily. ⚠ Note The “80% of RAG applications choose PostgreSQL” figure in the original document is directionally supported by practitioner surveys but an exact citation could not be independently verified at time of publication.

3. Microsoft’s Vote: Azure HorizonDB

If any single event crystallizes the shift in database industry alignment, it is Microsoft’s announcement at Build 2026. ✓ Verified

On June 2, 2026, on the opening day of Microsoft Build in San Francisco, Microsoft put Azure HorizonDB into public preview. The product had been first announced at Ignite in December 2025. HorizonDB is a fully managed, PostgreSQL-compatible database rebuilt from the ground up for agentic AI workloads. It supports auto-scaling storage up to 128 TB, horizontal compute scaling to 3,072 vCores, multi-region commit latency under one millisecond, and native DiskANN vector indexes with in-database AI model invocation via the azure_ai extension.

The announcement was authored by Affan Dar, Vice President of Engineering for PostgreSQL at Microsoft, and Charles Feddersen, Partner Director of Program Management for PostgreSQL — not Shireesh Thota as stated in some earlier drafts of reporting on the announcement. ⚠ Correction

The significance of HorizonDB is what it represents rather than what it does. Microsoft already operates Azure SQL Database, Azure SQL Managed Instance, and Azure Database for PostgreSQL Flexible Server. When it needed to build the database for the AI-agent era from scratch, it chose PostgreSQL as its protocol foundation — not SQL Server.

That choice is, as the original analysis puts it, a vote on ecosystem direction rather than a technical preference. Microsoft confirmed at Build that HorizonDB is already powering Web IQ — an AI grounding API rebuilt from Bing that provides the factual retrieval layer for both Microsoft Copilot and OpenAI’s ChatGPT. PostgreSQL is, in that sense, already running beneath a significant share of global AI interactions.

4. Capital Has Spoken: The PostgreSQL Acquisition Wave

The ecosystem signal does not stop at Microsoft. In 2025, three separate capital events — each independently verified — placed over $1.25 billion into PostgreSQL-native companies.

Databricks announced its acquisition of Neon in May 2025 for approximately $1 billion. Within weeks, Snowflake announced its own PostgreSQL acquisition — Crunchy Data — for $250 million. The two deals, from fierce rivals in the data warehousing market, arriving within a month of each other, reflect an industry-wide recognition that the OLTP future belongs to PostgreSQL.

Supabase’s April 2025 Series D — $200 million at a $2 billion valuation, led by Accel — confirmed the same thesis from the developer tools angle. Supabase reported $70 million in annualized recurring revenue by late 2025, making it one of the fastest-growing database infrastructure companies in history.

5. What Comes Next

The confluence of forces described above — ranking trajectory, architectural fit, AI integration, and concentrated capital — points in one direction. By late 2026, PostgreSQL is likely to hold the third position in global database popularity for the first time in the history of the DB-Engines index.

More significantly, the question of which database underpins the AI era appears to be settling. The “relational + vector” hybrid model that PostgreSQL enables through pgvector has, in practical terms, defeated the pure-play vector database as a default architecture for most teams. Dedicated vector databases remain relevant at extreme scale, but for the vast majority of AI-integrated applications, PostgreSQL handles both dimensions in a single, familiar system.

Microsoft voted with HorizonDB. Databricks and Snowflake voted with acquisitions. Developers voted with 100 million pgvector downloads. The database era of the 2020s is beginning to look a great deal like the PostgreSQL era.