Not always true. There are some very interesting projects at Google, Apple, etc. that can only be done by companies capable of throwing hundreds of
Millions/billions of dollars at problems with uncertain payoffs
That is why I said "almost always done" and not "always done".
Of course, a lot of that is subjective, as it depends on what you find interesting. I don't personally tend to find large-scale projects very interesting because they tend to be worked on by large teams, rendering most individual roles to something much more narrow. But I do know plenty of devs who get very excited about large-scale projects. Vive la différence!
Small companies make it possible, large companies make it economical.