It happens because of their internal promotion culture. You get a promotion much faster if you make something new than if you improve something old.
A good company has both kinds of developers, ideally more of the improve type than the build new type.
Google’s management failure is that the optimize for the “build new” engineer and actually punish the “improve” engineer by stalling their career.
Apple tends to do this well. Messages feels like one continuous evolution from the basic SMS app on the first iPhone. When iChat moved to Messages a couple features were lost, but the important ones are back now. It was a new name, but it felt more like a unification between iOS and macOS vs just thinking a new app would help them get more users.
If someone went from a v1 iPhone to an iPhone 13, the new one would have dramatically more features and capabilities, yet it will still feel familiar. There is a lot of value in that. Engineers can "build new" within an existing framework if the foundation is solid.
When the "build new" dev make something new it's up to management to decide if it should be a feature of the existing service or if it's foundational enough to require a whole new app. That choice to build new should not be taken lightly, as migrating millions of users is not a trivial task and they will lose people along the way.
Google Wave was the perfect example of that. Wave would have been amazing if it was integrated into Gmail. Like Apple did with iMessage, if a Gmail user emailed another Gmail user (or group of users) it would turn into a Wave and get a bunch of extra features. This would have provided value, instantly given it millions of users, and encouraged current users to get their friends and co-workers on Gmail to use Waves. Instead, they made it it's own invite-only service (trying to recreate the Gmail magic) and it was a failure due to lack of users.
This is a horrible strategy and they are going to continue shooting themselves in the foot if they don't pivot.