Let me tell you a little secret: streaming is also downloading. If this data were not downloaded, how would your computer know what audio to play?
The major difference between streaming and saving a file locally is that streaming is defective by design, so that what you download is single use only. But streaming is nothing special, the exact same zeroes and ones flow through the internet tubes to your computer whether you’re using yt-dlp to save a video file or you’re watching it using the crappy video player on the YouTube website.
The bottom line is: if you want to watch a 3 GiB YouTube video, you have to download 3 GiB, whether you save it locally as a file or the software used to play it forces you to delete that data once you’ve finished “consuming the content”.
In fact, if you want to conserve bandwidth and download less, it makes sense to save what download so you can play it again the next day, instead of “streaming” it, which means downloading it in its entirety every time you play it.