We're currently in the process of choosing an FPGA for our new gizmo. Our project will have a non-negligible cost and should last us for a long time.
Now, with each new FPGA of a particular line (say, Xilinx Virtex) being more powerful than its predecessors (the Virtex 7 vs. the Virtx 6 vs. the Virtex 5 etc), and since we'd like to choose an FPGA which is as powerful as possible to allow for future enhancements, we're tempted to choose the latest-and-greatest chip that we can get.
On the other hand, new chips and their toolchains tend to have their teething problems, and we certainly don't want to have any hassle from that side.
What's your rule of thunb - how old must an FPGA be so that it can be considered mature enough for a project? Or, how else would you gauge the fitness of an FPGA? By numbers of chips sold, or by the number of high-wire projects that use it?