What Microsoft Didn't Say at Build
The confetti was still on the floor when analysts started asking uncomfortable questions. Microsoft's Build conference had put its AI agent Scout front and center — a system designed to observe how work flows and act without being told. But the company's own flagship database product, SQL Server, was conspicuously absent from the conversation.
That silence might seem minor in isolation. But layered on top of other signals — the departure of Rohan Kumar, a longtime champion of SQL Server within Microsoft, and the fact that Arun Ulag now oversees both Azure Data and Fabric analytics alongside a growing portfolio of open-source database services — the pattern became impossible to ignore. Compounding the unease: Microsoft quietly updated its terms of service to let customers carry their SQL Server licenses over to Amazon's RDS platform, sidestepping double licensing through a bring-your-own-installation-media provision.
The message from the C-suite was clear: SQL Server is no longer the priority it once was.
The Analysts' Frustration
Andrew Snodgrass, research vice president at Directions on Microsoft, put it bluntly: "I don't think it is a priority. With Kumar leaving, that's become very evident. I'm afraid Microsoft are going to leave it languishing."
His concerns trace back to the 2022 release, which shipped with a suite of Azure integration features that analysts felt were driven more by marketing strategy than genuine customer demand. The 2025 edition did earn some respect for adding vector search — a capability that had already been standard in PostgreSQL, MongoDB, and Oracle for years.
Snodgrass noted the irony of Ulag's Build keynote: "He stood up there and talked about all the new stuff — highlights of the database news were HorizonDB, a PostgreSQL service with scale-out capabilities. There was no news about SQL Server, which was stunning, because SQL Server 2025 just came out at the end of last year with AI vector search, one of the greatest additions I've seen in ten years."
His verdict? "It's not that I'm angry at Microsoft for what they've done to SQL Server. I'm just disappointed."
The Revenue That Won't Disappear
Adam Ronthal, vice president analyst at Gartner, offered a more pragmatic reading of Microsoft's calculus. The company pulled in approximately $15 billion from the on-premises database market — a figure dominated by SQL Server. With a 33% share of that on-prem market (second only to Oracle's roughly 40%) and growth running around 8% in 2025, the financial incentive to maintain the product is enormous.
"There is no way that Microsoft will walk away from that kind of revenue," Ronthal said.
Beyond direct sales, SQL Server customers represent a natural upgrade path into Azure SQL and Microsoft Fabric. The company envisions a unified ecosystem where data, AI, security, and governance converge — but the practical reality of migrating T-SQL-heavy applications to managed services remains uneven, depending on how deeply an application relies on specific database configurations.
The Cloud Reality Check
The cloud database market tells a different story. PostgreSQL has emerged as the de facto standard for relational databases in the cloud, according to Ronthal. Startups like Cockroach Labs, Yugabyte, and pgEdge have all built distributed databases on PostgreSQL-compatible APIs, giving enterprises a hedge against vendor lock-in.
Microsoft's response has been twofold: HorizonDB, its own distributed PostgreSQL offering, and Azure Database for PostgreSQL. The company is also folding operational databases — Cosmos DB, Azure SQL, Postgres capabilities — under the Fabric umbrella. "Fabric is a core component of AI," Ronthal explained, "and the growth for Microsoft is largely going to be driven by Fabric adoption."
But the numbers reveal a significant gap. In 2025, AWS generated roughly $37 billion in cloud database revenue compared to Microsoft's approximately $18.3 billion — a nearly two-to-one disadvantage that underscores how far behind the Redmond giant still is in cloud-native database dominance.
The BYOL Paradox
Microsoft's decision to allow SQL Server customers to bring their licenses to AWS RDS creates an interesting strategic tension. On one hand, it means Microsoft loses potential Azure migration revenue — a customer running SQL Server on Amazon's infrastructure isn't contributing to the Azure database ecosystem.
On the other hand, Ronthal pointed out that "they don't lose a SQL Server customer, and that's probably more important." The licensing revenue continues flowing even if the compute environment shifts. It's a rational calculation: preserve the license income stream while accepting that the underlying infrastructure may not be Microsoft's.
This arrangement also reflects a broader industry trend — as open-source alternatives gain traction, proprietary database vendors are increasingly forced to make their products portable across cloud providers rather than demanding exclusive deployment.
The Portfolio Problem
Ronthal identified a structural challenge that could complicate matters further: Microsoft currently offers multiple ways to run SQL Server, each with different compatibility profiles. Azure SQL, managed instances, and SQL Server running in virtual machines all sit somewhere along a spectrum of control and convenience.
"You've got Azure SQL, managed instances, SQL Server in VMs," he noted. "These provide slightly different levels of compatibility with what you might be doing in the on-prem world, and right now the fact that there are multiple options actually makes it difficult for end users to figure out what to do. I would love to see Microsoft make it more unified and easier for people to consume."
This fragmentation could become a liability as customers evaluate their migration paths, particularly if Microsoft continues to deprioritize investment in the core SQL Server engine while pushing users toward managed alternatives.
The Long View
Despite the strategic ambiguity surrounding SQL Server, Microsoft's track record in the database market remains impressive. Of the major vendors — Oracle, IBM, Microsoft, and SAP — only Microsoft has expanded its market share over the past 15 years. The company has demonstrated an ability to navigate market transitions, balancing legacy product support with cloud innovation.
For SQL Server customers, the 2025 version carries support through 2036 — a decade of runway that provides ample time to evaluate options and plan migrations. Whether Microsoft's dual strategy of defending on-prem revenue while investing heavily in PostgreSQL and cloud-native alternatives ultimately serves its database customers well remains an open question. The company has proven it can survive market shifts; the question is whether SQL Server users will benefit from that survival or simply be collateral in a broader strategic realignment.