But the used Snapdragon 801 (8974-AA) SoC was released in Q3 2014 (and used in the Blackberry Passport since September 2014), nearly 3 years ago. The SoC being the most deciding component for obsolescence, for the purpose of this discussion I would side with edanto that the hardware is nearly three years old despite the product being assembled significantly later.
And I think herin lies an important lesson for Fairphone: the economic lifespan of an SoC is approximately two years. SoCs and production processes still evolve fast enough (in terms of perf and perf/watt) for parties like Qualcomm to keep releasing new SoCs every year or more often. These SoCs will inevitably be incompatible with earlier chips. With a bit of stretch, you can get OS upgrades for these two years and provide a well supported phone for 3-4 years, but only if you manage to release your phone at the same time as the SoC. Parties like Fairphone don’t have enough volume or engineering momentum to pull this off; it means you’d have to develop the mainboard for the phone from a Qualcomm engineering sample in the few weeks long before final production of the SoC.
Another point I’d like to stress (again ;-)) is that upstream kernels matter! The more the Android kernel diverges from upstream, the more effort needs to be made for an Android upgrade, and thus the less likely it is to happen. On the contrary, if drivers and CPU support live upstream (and Qualcomm is doing quite well in making this happen), the community takes care of this as frameworks evolve.
Hopefully for the hypothetical FP3 you guys can start with an SoC that comes fresh from the presses and make real effort to push for a kernel that is as close to upstream as possible!