LinkedIn is doing that thing again. 'Is SQL Server Dead?' posts are getting traction, the comment sections are lit, and I'm sure a lot of managers out there are forwarding one of them to a DBA with a question mark and no context. It happens every few years. NoSQL was going to kill it. The cloud was going to kill it. Now it's AI and Fabric. I've been doing this for 28 years. Let me save you some time.
No. SQL Server is not dead or even dying. But something is changing — and we need to understand what.
The Market Signal
Start with something real. Brent Ozar — who has been in the industry longer than many of those LinkedIn posters — wrote in February that his consulting pipeline nearly shut off in December. Not because SQL Server is irrelevant. But because clients started telling him: 'For 2026, any time we want to buy something, build something, or hire someone, we're going to try AI first and see what happens.'
That's not a LinkedIn hot take. That's a working consultant describing what his clients are actually saying. Pay attention to that.
What Microsoft Said in Atlanta
Eight thousand people descended on Atlanta in March for FabCon and SQLCon — the first time Microsoft's SQL community conference was co-located with the Fabric conference. The headline message from the keynote: 'SQL Server is not being replaced. It is being elevated.'
I believe them. SQL Server 2025 is reportedly growing faster than any previous version. The product is genuinely good. Microsoft isn't walking away from it.
But read the fine print. Every piece of the 'unified platform' story — the Database Hub, the migration experience, the single pane of glass across your entire database estate — runs through Azure Arc. If you want a seat at that table, your on-prem SQL Server needs to be Arc-enabled. Phone home to Azure. That's not a knock. It's just the architecture. Know what you're agreeing to before you agree to it.
What AI Actually Does to This Job
I've spent the last several months testing AI against real client workloads. Six posts on sp_BlitzCache and ChatGPT. The sp_BlitzIndex @AI parameter. Stored procedure refactoring. The MCP server. I'm not theorizing — I have results.
Here's what I found: AI handles the easy stuff faster than I do. Index suggestions on a clean, well-documented table? Great. Obvious query rewrites? Fine. Spotting missing statistics? Sure. If your job consists entirely of those things, yes, you should be paying attention to the market signal above.
But Brent said something in January that I think is exactly right: people doing mission-critical, secure, accurate database development on large existing databases will still struggle in 2026 due to undocumented databases and bad tooling. AI sees the query. The DBA knows the story. The DBA knows which index exists because of an outage three years ago. The DBA knows which stored procedure cannot be rewritten for political reasons. And they know why the schema is what it is — and why it cannot be changed.
That institutional knowledge is not in a prompt. It is not in a system message. It lives in the head of the person who has been sitting in that environment, and no amount of context window is going to replace that.
The On-Prem Majority
Here's the thing about FabCon: the people in that room skew heavily toward cloud-forward. They're already moving. They're already building on Fabric. They are not your clients, and very likely they are not you.
Observers at the Atlanta event noted that the SQL Server on-prem base was noticeably more visible than at FabCon Vienna in 2025, where most attendees seemed already committed to Azure SQL or cloud-first patterns. That's very telling. The on-prem majority showed up in Atlanta and made itself known — because it isn't going anywhere anytime soon.
The surveys back this up. Most production SQL Server environments are running hybrid: a mix of on-prem, rented data center, and cloud. That picture does not change overnight. And it certainly doesn't change because someone posted something on LinkedIn.
So What Is Actually Changing
The job description is shifting. Not disappearing — shifting. The shops that are going to feel it first are the ones where the DBA's value is mostly procedural: run the index job, apply the patch, restore the backup. If that's the bulk of what you do, AI is coming for your job — and it's coming faster than most people are ready for.
The shops that will be fine are the ones with a human who understands the environment at a level no tool can replicate. The one who can look at a blocking chain and know, without running a single query, which application is probably causing it. The one who remembers the last time someone tried to change that setting and what happened. The one who knows the business, not just the database.
That's not a description of someone who should be afraid of AI. It's a description of someone who should be using it — which is exactly what I've been doing. I stopped worrying about whether AI belongs in my toolkit. It does. It gets me where I need to go faster.
SQL Server has been declared "dying" or "replaced" since the rise of NoSQL a decade ago, yet it remains one of the top four most used databases worldwide in 2025–2026, alongside MySQL, PostgreSQL, and SQLite.
SQL Server has been around for 37 years. I'm betting it will outlast this conversation.
More to Read
The Tech Consulting Market Isn't Looking Good — Brent Ozar
Database Development with AI in 2026 — Brent Ozar
FabCon and SQLCon 2026: Unifying Databases and Fabric — Microsoft Azure Blog
SQL Server at FabCon and SQLCon 2026 — John Deardurff
Event Debrief: FabCon / SQLCon Atlanta 2026 — WhereScape
sp_BlitzCache + ChatGPT Series — sqlfingers.com
No comments:
Post a Comment