Is Microsoft still the right choice for a new server project?

For a genuinely new project, PostgreSQL on Linux is usually the first option I'd evaluate. It avoids licensing costs, is mature and scalable, and is well supported by most cloud providers. SQL Server and Windows can still make sense when you already have Microsoft infrastructure, specialized staff, or applications that depend on them.

The answer can also depend on scale and licensing details. A small SQL Server deployment may be affordable, while enterprise editions and large cloud instances become extremely expensive. Compare the full cost—including support, staff time, migration effort, backups, and operations—rather than looking only at the database license.

The main question is less about which technology is universally best and more about your existing constraints. If the business already relies on a Microsoft-heavy environment, has legacy .NET Framework applications, or needs tight interoperability with an existing ERP, switching can take a lot of time and introduce risk. In that situation, paying for Microsoft may be cheaper than rewriting everything.

Cloud hosting doesn't automatically make Microsoft the better choice. Managed databases and servers save you from handling hardware, networking, backups, redundancy, patching, and cooling, but the licensing can be very expensive—especially for SQL Server and Oracle. Managed PostgreSQL or MySQL is often considerably cheaper, while still providing much of the operational convenience.

There are also legitimate reasons to stay with Windows and SQL Server: your team may already be productive with them, clients may expect vendor-backed support, and accountability can be easier when a major vendor stands behind the platform. If learning and operating a different stack would delay the project significantly, the license cost may be a reasonable tradeoff.