It seems to me that this decision to use thin clients would have three aspects - cost, developer productivity and developer morale.
Developer morale is easiest to estimate. Most developers(and other people too) love having big, fast machine. I don't think anybody would like switching to thin clients.
The cost would depend on how many clients per server you support. The professor in my virtualization course told me that modern servers can easily support 140 users(This would of course change with time and the type of servers you use). However, this for typical workers who use computers as productivity tools. Building typically requires 100% CPU, while average PC load otherwise is usually in 5-20% range. So depending on how large your build cycles are, the number of clients per server you could handle would change. (Larger build cycles -> more parallel builds -> more load on server). Make sure you make management understand that supporting developers using thin client is more costly than for non-developers. Note that IT cost may be also reduced by switching to the thin client model.
I can't really talk about how productivity would be affected, except to guess that productivity should not be reduced as long as your servers are fast.