I read the paper your attachment is from, however it does show the downsides of assumptions. As far as we're aware between now and 2020 at least exactly zero of the G.fast 'investment' will be reusable for anything else. I guess that scenario assumed BT would put fibre deeper into the network. Perhaps naive to think BT would spend more than they absolutely had to initially to keep the politicians off their back.
Be interesting to see what impact this has on the maths. Obviously it dramatically reduces the cost of the deployment, however it will also shorten its lifespan or necessitate further upgrades.
This is assuming that said further upgrades would be done on a commercial basis, which given the coverage BT can get, albeit at barely higher than FTTC rates for some, it may not be.