As it turns out, chip fabrication is very complicated. Not as easy as commenting on an internet site. (Who would have thought). AMD/Global foundry failed to deliver 7nm, Intel had a lot of delays with 10nm/Intel 7 and delivered a sub par product. Samsung's 5nm LPP is notoriously bad. Now it is TSMC's turn.
TSMC N5 and its derivative N4 is one of a kind, the exception rather than the norm. It had good yields and low leakage. We can't expect them to consistently over perform, just like we can't exist Olympic champions to win every year.
For people who claim Apple should wait for N3E - did you know R&D for fabrication is costly? If no one is putting in orders then why would TSMC to keep throwing money into the fire pit? They are not philanthropists or volunteers for science. They want to make money, which means customers foot the bill. If no one buys N3B then N3E will take longer to come out and we will be stuck with N5 for longer (not that I'm complaining because N5 is good).
My prediction is we will be stagnant for a while. Expect the current gen to be good efficiency and the next few gen to be mediocre improvements. 8+ gen 2 and A15 is still the king for the next few years. They may still improve raw performance by overclocking w/o improving efficiency, but all it means is you'll have hotter products with less battery life just to claim some benchmark numbers (888 and 8 gen 1 for example).