What's new in SQL Server 2025 and Azure SQL Databases - Christian Peeters - NDC Oslo 2026

Item: What's new in SQL Server 2025 and Azure SQL Databases - Christian Peeters - NDC Oslo 2026
Rating: 4
Author: PostSingularity Institute

Title history

What's new in SQL Server 2025 and Azure SQL Databases - Christian Peeters - NDC Oslo 2026 — observed on 20261028-0836 UTC (YouTube)

by Christian Peeters · View the channel on PSI

Applied Sciences & Engineering Computing & Cybersecurity UNADatabase design and theory UNARRelational databases

🎙 Christian Peeters 👥 230K 📅 October 28, 2026 ⏱ 60 min 👁 4 📄 tutorial 🧭 2026-10-28

Available in: English (current) Français

Keywords

SQL Server 2025Azure SQLvector searchembeddingRAG

Summary

The speaker introduces the new capabilities of SQL Server 2025 and Azure SQL databases, focusing on AI-driven features that transform the database into an active component of intelligent applications. He begins with a brief history of SQL Server, noting the progression from relational tables to XML, JSON, and graph support, culminating in the 2025 release’s native vector data type and vector search functions. The core of the talk explains how embeddings represent the meaning of text as high-dimensional vectors, and how cosine similarity and vector distance allow semantic search without exact keyword matching. Using a live demo with a fictional musical instrument shop, the speaker demonstrates manually implementing embeddings with an external model and then shows how SQL Server 2025 simplifies this with a built-in vector type, the VECTOR_DISTANCE function, and support for external AI models via credentials and model definitions. He highlights the performance benefits of vector indexes and the difference between exact KNN and approximate ANN search. The session also covers related T-SQL enhancements such as REST endpoint calls, regular expressions, fuzzy string matching, and improved JSON processing, emphasizing developer productivity. The talk concludes with a brief comparison between SQL Server on-premises and Azure SQL, noting that Azure already supports updates on indexed vector columns.

207 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of the information is high for database developers and architects interested in integrating AI capabilities directly into SQL Server. The speaker provides a clear, progressive explanation of vector semantics and demonstrates how new features reduce complexity compared to manual implementations. His argumentation is solid, relying on practical demos and a logical structure that builds from fundamental concepts to advanced usage. He addresses trade-offs, such as the inability to update indexed columns in SQL Server 2025 (fixed in Azure), which enhances credibility. However, the talk lacks quantitative benchmarks or independent validation, and some performance claims are based on anecdotal demonstrations.

Scientific Rigor, Source Quality, Title Accuracy

The rigor is acceptable for a conference talk; the speaker references his own experience and the product documentation indirectly. He mentions using Azure samples from GitHub for the manual vector search code, but no explicit citations are provided beyond the conference links. The title accurately reflects the content, and the session aligns with the stated focus on AI and developer productivity. The lack of formal sources reduces the scientific rigor, but the practical nature of the demo compensates somewhat. The conference itself (NDC) is reputable, and the speaker’s role as a principal cloud architect supports his expertise.

213 words

Title / Content Match

The title accurately reflects the content, which focuses on new features in SQL Server 2025 and Azure SQL, particularly around vector support, AI integration, and developer productivity.

Quality & Reliability

7/10

The talk presents concrete demonstrations of SQL Server 2025 and Azure SQL features, backed by practical examples from a principal cloud architect. However, it is a vendor-agnostic conference talk rather than a peer-reviewed publication, and some claims about performance and capabilities lack formal benchmarking or citation.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

NDC Conferences — Conference organizer and official event page
NDC Oslo — Specific conference event page for NDC Oslo

External References

ndcoslo.com

Contribution & Novelties

The presentation showcases the new vector data type and vector search capabilities in SQL Server 2025 and Azure SQL, which enable native semantic search and RAG scenarios directly within the database. By integrating external AI models, SQL Server can now generate embeddings without custom application code, significantly reducing complexity for developers. The talk also highlights improvements in T-SQL for developer productivity, such as REST endpoint integration and advanced string processing. This represents a major shift from storing data to actively understanding and processing it within the database engine.

Pour aller plus loin :

Vector database — Explains the concept of storing and searching high-dimensional vectors.
Retrieval-augmented generation — Overview of RAG, a key use case demonstrated in the talk.
Word embedding — Background on how embeddings encode semantic meaning into vectors.

131 words

Radar Profile

The radar profile shows a balanced score across information quality, technical depth, and reliability, with slightly lower reliability due to the absence of formal citations. The high quantity of information and technical level reflect a dense, hands-on presentation aimed at database professionals.

Reliability 7/10