There are no truly free compute resources (galaxy may work if your can work within its storage limits or https://cyverse.org/ may be another option) that would allow one to work with large sequence datasets. Generally people will have access to institutional resources to do this kind of work. It is possible to subscribe to cloud compute (if you are able). One can't analyze datasets by streaming from SRA.

If you eventually intend to create datasets of your own in future please make sure you can arrange the compute resources beforehand before you invest time/effort into creating those datasets.

Perhaps you (or your mentor) can try and collaborate with another institution/lab, which may have the compute resources available.

The truth is, as you progress, you’ll need more and more storage space.

One strategy I’ve found works is to use Google Drive + Google Colab Pro. This way, you can download your reads directly to Google Drive and link it to Google Colab. It’s slow, but it works.

However, I’d recommend you buy a portable SSD straight away and store your files on it. Next, you’ll need to access an HPC or institutional server.

I hope it helps