June 2025 was a heavy month for data infrastructure, with two large acquisitions, a major Microsoft Fabric update, and a notable architecture announcement from Netflix. Here are the developments worth tracking.

Databricks acquires Neon for ~$1 billion

Databricks agreed to acquire Neon, a serverless PostgreSQL company, for roughly $1 billion. The stated rationale centered on agent-driven workloads: by Neon's own figures, more than 80% of the databases provisioned on its platform were created programmatically by automated agents rather than by humans. Serverless Postgres that can be spun up and torn down on demand fits the pattern of AI systems provisioning their own short-lived storage.

Snowflake acquires Crunchy Data for ~$250 million

Snowflake acquired Crunchy Data for around $250 million. Crunchy specializes in enterprise- and government-grade PostgreSQL, including FedRAMP-oriented deployments. The deal gives Snowflake a hardened Postgres offering and signals continued convergence between analytical platforms and operational Postgres.

Microsoft Fabric: Variable Libraries and Materialized Lake Views

Microsoft shipped a sizable Fabric update. Two items stood out:

• Variable Libraries - reusable, parameterized values shared across Fabric items, reducing duplicated configuration and environment-specific edits when promoting between dev, test, and production.
• Materialized Lake Views - declaratively defined, automatically maintained views over lakehouse data, with built-in refresh and lineage so common aggregations don't have to be hand-managed as pipelines.

Netflix Unified Data Architecture (UDA)

Netflix detailed its Unified Data Architecture (UDA), a model where a concept such as "movie" or "member" is defined once and that definition propagates across services, pipelines, and tooling. The goal is a single source of truth for entity semantics so teams don't re-implement the same definitions inconsistently across systems, cutting drift and integration bugs as features ship.

The through-line

The acquisitions and product releases point in the same direction: PostgreSQL is increasingly the operational layer next to analytical platforms, and a growing share of database provisioning is driven by automated agents rather than people. The Fabric and Netflix work reflects a parallel push toward defining data assets once and maintaining them declaratively rather than through hand-built pipelines.