They don't do the same thing.

The dict comprehension requires a single key:value pair per loop. It accumulates values into a single dict. Try this:

    d = {}
    for key, value in items:
        print(id(d))
        d[key] = value

The ID doesn't change because it is the same dict each time.

Unpacking a dict doesn't produce a single key:value pair, except maybe by accident, so it is not usable in a dict comprehension.

Your second example doesn't modify a single dict, it **replaces** it with a new dict each time.

    d = {}
    for sub_dict in super_dict.values():
        print(id(d))
        d = { **d, **sub_dict }

The IDs will change through the loop as d gets replaced with a new dict each time. So this is not equivalent to a comprehension.

Also, the second would also be very inefficient. It unpacks the existing dict, then packs the values into a new dict, then unpacks it again, then repacks it into yet another dict, and so on.

Better:

    d = {}
    for sub_dict in super_dict.values():
        d.update(sub_dict)

But that's not equivalent to a dict comprehension either.