Optimizing BZ2 decompression code for big data set

Question

I have the following fully working code that:

1. Walks through all sub-folders in a given directory;
2. Captures dirpath and filename for the compressed file (i.e., Dir/*.json.bz2);
3. Creates a decompressed file with the same dirpath and new file ending (i.e., Dir/*.json). Notably, each newly created decompressed file has a unique numbered filename (i.e., 1.json, 2.json, ..., n.json);
4. Decompresses BZ2 archives and writes them on disk.

I tested the code using 1,380 BZ2 archives (~1.5GB) and it takes substantial time to process them. Considering that my goal is to decompress 40,000 BZ2 archives (~45Gb) per run, please suggest if there is a plausible way to optimize the code in order to increase its speed. Thanks!

import os

file_counter = 0
for dirpath, dirname, files in os.walk('/Users/mymac/Documents/Dir'):
    for filename in files:
        file_counter += 1
        if filename.endswith('.json.bz2'):
            filepath = os.path.join(dirpath, filename)
            newfilepath = os.path.join(dirpath, "{0}.json".format(file_counter))
            with open(newfilepath, 'wb') as new_file, bz2.BZ2File(filepath, 'rb') as file:
                for data in iter(lambda : file.read(100 * 1024), b''):
                    new_file.write(data)

Additionally, I am curious which option will be more efficient: (A) run the code and then delete all processed BZ2's, or (B) add some lines in the code and delete each decompressed BZ2 while running the code?