What do you do when a programmer tells you "this can't be done" (perhaps "... with these resources", including time)?
You're sure to find many others who say "no big problem!"
Which do you believe? The wrong answer can be very expensive either way.
There's a reason why very few '7-80s superminicomputer companies survived the advent of multiprocessors and trying to get their operating system(s) to work reliably with them.
In that case, it was "this will be hard" vs. "not too hard". In most cases, the B team got the nod, and all was well ... until customers with machines in the field started complaining how their shiny new systems would sporadically crash. And the B team couldn't find the problems (or at least very quickly), since among other things they hadn't invested enough in instrumentation and so on. Some companies were lucky enough to get a chance to hire third team who could do it, but I think in most if not all of those cases the company was too injured and too late with a good solution to survive.
ADDED: I should probably qualify this with "it depends a lot on how hard the problem is". I'm used to the domain of hard technical problems because that's mostly how I made my living before retiring. And one advantage of them is the barrier to entry they pose to competitors.
After all, we're here because the problem of developing a really good (easy to use and so on) web shop development system is hard, something where using secret alien technology (http://lispers.org/) can make a big difference, and did.