# The Modern Data Stack in 2025
The definition of the "Modern Data Stack" (MDS) has shifted. If 2020 was about ELT and centralized warehousing, 2025 is about **Data Contracts, Semantic Layers, and AI Orchestration.**
1. Governance as Code
We've moved beyond documentation into active enforcement. Data contracts are now the standard for ensuring upstream changes don't break downstream analytics.
2. The Rise of the Semantic Layer
Tools like dbt Semantic Layer and Cube are allowing organizations to define metrics once and use them across all BI tools, ensuring a single source of truth.
3. Real-time becomes Default
With technologies like ClickHouse and Apache Pinot, the barrier between analytical and operational data is blurring. Real-time dashboards are no longer a luxury; they are expected.
4. AI-Native Infrastructure
The modern warehouse is now also a vector store. Snowpark and BigQuery ML are bringing the compute to the data, rather than moving data to the compute.