Semver (or really, the concept of a single unified version number) makes it difficult to compare small versus large projects. Large projects have less information in the version number.
Let's say you have N small projects, each of which gets a major version bump once every two years. Now, combine them into a single umbrella project. If you have 24 of them and they don't coordinate, you'll end up bumping the major version once a month (on average), even though the code didn't really change any more than before.
To somewhat improve on this, you can coordinate release dates, and you get a regular major version bump, the times when it's always true that some things changed, even though most things didn't. Either way, the version bump itself doesn't tell you as much compared to having separate projects.
Let's say you have N small projects, each of which gets a major version bump once every two years. Now, combine them into a single umbrella project. If you have 24 of them and they don't coordinate, you'll end up bumping the major version once a month (on average), even though the code didn't really change any more than before.
To somewhat improve on this, you can coordinate release dates, and you get a regular major version bump, the times when it's always true that some things changed, even though most things didn't. Either way, the version bump itself doesn't tell you as much compared to having separate projects.