They already released all their macbooks and latest iphone on N3B which is the worst-yielding 3nm from TSMC. I doubt yields are the issue here.

It's suspected that the fast release for M4 is so TSMC can move away from the horrible-yielding N3B to N3E.

Unfortunately, N3E is less dense. Paired with a couple more little cores, an increase in little core size, 2x larger NPU, etc, I'd guess that while M3 seems to be around 145mm2, this one is going to be quite a bit larger (160mm2?) with the size hopefully being offset by decreased wafer costs.

I think this is the most likely explanation. Lower volume for the given product matches supply better, and since it's clocked down and has a lower target for GPU cores it has better yields.