I think there are a couple reasons for this.
First, historically, Apple Watch development was difficult and very constrained. The framework had a lot of limitations and, on early watch models, third party apps were very slow compared to native apps (which themselves were not very speedy).[1]
This is changing now as the hardware gets better and Apple is opening up more powerful and easier APIs to third-party developers, but it will take a while for third party apps to catch up. And the situation is still not as good as it is on iOS.
The other reason is that there's a small number of app concepts that even make sense on the watch. The watch is very good at a few things: quick, glanceable interactions (like notifications) and fitness tracking. Apple suggests that apps should "support fast interactions and focus on the content that users care about most", and that "interactions with the Apple Watch are measured in seconds." [2]
A great deal of apps simply don't fit this category at all. Many existing iOS apps expose some basic watch functionality, but many won't, because it just doesn't make sense. Consider how willing you would be to hold your wrist up in front of you for 30+ seconds.
[1] https://marco.org/2018/02/26/watchkit-baby-apps
[2] https://developer.apple.com/design/human-interface-guideline...