I'd have to take the alternate view to Surfinjo once again. I exist in a place where there is NEVER enough CPU power, but I am limited by both up front cost and running cost (energy = heat). I know I am not alone in this, even if we're not the mass market. But this isn't aimed at the mass market. It is pointless judging it for such a use since there are other more suitable products to fit that role.
So, back onto Broadwell-E, it will always come down to the value proposition. At which point do you say enough is enough? When building a new system some months ago, I had to ask myself, is it better it go for Skylake which had just come out, or go for Haswell-E? It was a gamble but I went the mainstream route of Skylake, and I think that is the slightly better long term option. To me the factor isn't simply how many cores you get, but you need to balance the rest of the system with it. How well does your workload scale with cores? Will you run into ram bandwidth limitations, because I sure do. On that note, I wonder if Broadwell-E will include eDRAM like the higher end consumer versions? That could help mitigate ram bandwidth limitations, although with Haswell-E as precedent, it'll likely have quad channel ram anyway which is a help over trying to get dual channels running stupidly fast.
Basically for me it is impossible to make any decisions without hard facts. What is the actual core/clock configuration? How much cache at what level? Rated TDP?
I'm looking to build a 2nd overclocked Skylake rig... don't think it is worth waiting for Broadwell-E myself... the problem with all these -E skus is they're essentially branched off the Xeon line, which comes very late compared to the consumer chips.