That's my interpretation, based on reading a variety of BT statements alongside a number of analysis pieces. It might not be quite so black and white in the end, though.
And, IMHO, the biggest issue isn't about the telco side of things, but about humble power.
My understanding of the current FTTC rollout (in both the commercial and BDUK portions) is that power is one of the biggest problem areas. Even as standard (a one-of payment of around £1,000), the cost of power installation is one line item that makes or breaks a cabinet; excess costs can easily bring the cabinet out of the project - and this can happen if the power company passes on costs, eg for new transformers. As an example, 3 cabinets were found in one town in my county to need £90,000 to get power to them; they were dropped from the project until the LA used some "leverage" to persuade BT to include them.
The power problem extends to FTTRN, and is what has stalled progress there. The only existing/known rural node needed a separate power cabinet building for it.
If we assume that a G.Fast rollout will be plagued by the same issue, I think we are standing on safe ground.
So, having said that, it is easy to see that BT can probably manage to deploy a G.fast node at the existing FTTC site without worrying about the cost of power, as they can feed it off their existing one. However, any extra nodes will either need local power of their own, or will need "forward power" fed over spare copper from the current node, or will depend on "reverse power" fed from the end users.
Deploying extra nodes will become a trade-off between the distance away from end-users (fewer nodes, slower speeds) vs the distance away from the existing cabinet ("forward power" losses depending on length of copper) vs the maturity and acceptability of "reverse power" solutions.
Cabinet-only becomes a relatively easy no-brainer, while extra nodes need some finesse.
But there's an upper limit, financially, to all this: there's no point spending more (in any one area) than an FTTH rollout would cost.