Lower single core throughput with more cores is just always a bad bet. Existing software runs better on faster cores, people buy hardware to run existing software, and people write software to run well on hardware that other people already own.
The reason for AMD's resurgence right now is not that they have more cores, but that they have better cores. If they had even faster cores, and fewer of them per die, they'd be selling even better.
Well. I got the very first Ryzen model, the 1800X, because it had twice the amount of cores that Intel was selling, and they were just a few percent slower per core. If they had been 40% slower, I would have passed.
My most important workload - compiling C++ - is atypically parallel, but even there, single-core is important, too.
I think AMD was both lucky and good. They came out with a forward thinking design that could bring them back from the brink, but I'm not sure their stuff would have sold if Intel hadn't left them an opening. Most importantly was Intel's failure to execute on 10nm, global foundries 14nm wouldn't have compared as favorably to 10nm even with more cores. And since Intel was stubbornly refusing to sell more cores on anything but their expensive HEDT platform there was a market segment being neglected.
Zen was actually, in a way, a conservative design. They specifically mentioned "performance compatibility" with Intel - they couldn't be weirdly slow in some workloads because of an exotic design. You could say that they had the luxury of not having to aim for the moon because Intel was pretty much parked on the ground, with a failing new process and no ambition to increase core count. Against the original SV ethos, against Andy Grove, just plain old greed and complacency.
The reason for AMD's resurgence right now is not that they have more cores, but that they have better cores. If they had even faster cores, and fewer of them per die, they'd be selling even better.