Microsoft pushes SQL Server 2025 despite cloud‑first focus

Microsoft announced that SQL Server 2025 will ship with a suite of new capabilities aimed at on‑premises, hybrid and Azure‑VM deployments, underscoring the product’s continued strategic importance despite the industry’s cloud‑first narrative [1]. The move reflects Microsoft’s belief that large‑scale workloads still demand the performance, control and availability that only a traditional database engine can guarantee, and that the company’s broader data platform revenue—estimated at about $14 billion annually by Gartner—relies heavily on SQL Server and its Azure SQL services [1].

At the heart of the push is the “bring your model” AI integration, which lets customers run custom machine‑learning models directly inside the database, a feature designed to address emerging security concerns around AI workloads. Microsoft also points to the tight coupling between on‑premises and cloud services: many innovations debut in Azure SQL before migrating to the boxed product, and vice‑versa, ensuring a consistent T‑SQL experience across Fabric, Azure SQL Database and SQL Server itself [1]. This cross‑product code sharing means that investments in one area quickly ripple through the entire ecosystem, reinforcing Microsoft’s developer‑experience narrative.

Cost and performance remain key differentiators. While the cloud is often touted as cheaper, Joey D’Antoni notes that for workloads that consume large volumes of ultra‑fast storage and run at high utilization, on‑premises or Azure‑VM deployments can be more economical than PaaS offerings where compute and storage scale together [1]. He cites examples of customers achieving 13 million transactions per second on a single server—a performance level that drives many of the new features in SQL Server 2025. Moreover, the product’s mature high‑availability stack—availability groups, log shipping and failover cluster instances—offers out‑of‑the‑box resilience that open‑source databases on Linux still struggle to match, especially in mixed‑OS environments [1].

Licensing remains a pain point, with Enterprise Edition priced around $7,000 per CPU core plus software assurance, a cost that open‑source alternatives like PostgreSQL and MySQL can undercut [2]. Yet D’Antoni argues that the built‑in capabilities for backup, disaster recovery and high availability often justify the premium for enterprises that cannot afford the operational complexity of stitching together multiple third‑party tools.

The real question is whether Microsoft’s hybrid bet will keep SQL Server relevant as more workloads migrate to fully managed cloud services. If the new AI and performance features can deliver tangible productivity gains, the legacy product may remain a profitable pillar of Microsoft’s data revenue; if not, the high licensing bar could accelerate a shift toward cheaper, open‑source options.