Pretty lame story. It misses the whole shift in network architecture that V.fast needs to underpin it.
V.Fast does work from "a" cabinet, but isn't very likely to be the cabinets as we know them.
V.Fast will give high speeds in the 100-200 metre region, and is meant to tie in with a concept known as FTTdp, where the "cabinet" is located at the distribution point, is much smaller, and perhaps serves a few dozen properties.
In the current network architecture, perhaps only 20% of premises can be found within 200 metres of existing cabinets. Is there likely to be enough take-up within that portion to make it worth upgrading the whole cabinet? Especially as all those customers will be able to get 100Mbps vectored VDSL2 by then?
So V.fast is unlikely to happen with existing cabinets. It needs new cabinets, adding them at distribution points (either up telegraph poles, or placed underground), and being deeper into the network.
There's a UK-centric summary here:
There's another US-centric version of the same slides here, with more technical information:
It is still cheaper than full FTTH - probably - but the control of costs will need to be even tighter, or you might as well go for FTTP instead.
I suspect, though, that the decision about whether or not to ever deploy FTTdp and V.fast will depend on the take-up of FTTPoD.