And here's the rest.
It is a complicated picture of network optimisation, so you'd think @Ignition would have a good handle on this
C2) Future amendments to G.Fast
Is 500Mbps the end of G.Fast? I don't think so - it can offer more, but will do so only at shorter ranges.
If BT want to keep G.Fast at the core of a hyperfast strategy, not just an ultrafast one (and I think they do), they will need to gradually shorten those lines.
D) Driven by Bean Counters
Yes, the money will drive this.
Income
The "superfast" (largely FTTC) rollout ended up being a "just in time" rollout that rode the wave of HD IPTV, which has led to quite high takeup relative to BT's original expectations. But I don't think there is a similar "mass wave" willing to jump onto real ultrafast speeds. There might be demand from 5% of properties (reasonable), and it might be as high as 40% (I doubt it). The bean counters will have to figure out where this market is, to get the biggest step in takeup.
If BT targets a DPU at the PCP, and the 300m circle around it, then they will only be addressing people who can already achieve 80Mbps. What percentage will crave a jump to 300Mbps? 5%? 10%
If BT targets a DPU 400m away from the DPU, then they will be addressing people who can achieve 50Mbps-ish. Maybe you'll target more. Put your DPU further away, and you're likely to attract more custom.
A PCP-only strategy might hit the "homes passed" counter quickly, but may fail on the "takeup" counter.
Expenditure
The fibre runs will be much shorter this time, so will be less of the cost. Stepping 300-500m away from the cabinet might not be the hurdle that many people fear. However, power will be a primary issue - even when it is already on-site at the PCP.
If BT can power a DPU with a 240V supply, then this might be all they end up doing: one node at the PCP, on an extension lead
However, if BT have to go to the trouble of installing a power pillar, with 240V->50V conversion, then a single DPU at the PCP might not be a worthwhile return on that expenditure. If so, expect more DPUs inserted (say) 300-500m out from the PCP.
But that will only happen if power from the pillar can reach that 300-500m distance. I'm not sure.
On balance, from an income and expenditure perspective, I can see it being better to deploy more nodes than "just" at the PCP.
E) How big a market?
Actual speeds at VM tells us there is a market for top tier of only around 10%. And a "barely ultrafast" market for the next 30%, though some of these might be "reluctant ultrafast", having received a "free" upgrade from the speed they originally chose the middle tier for.
But, these comparisons pre-supposes that VM is a direct competitor to BT, and I'm not sure it is entirely. When you bring TV into consideration, there's a good chunk of VM - half perhaps - that get labelled as "premium TV" (and a bigger chunk of Sky). The premium bundles tend to include faster broadband as a matter of course. BT's TV offering isn't going to convert these people, so it isn't worth considering what they might do about broadband choices.
Back in the "BT-based DSL" market, less than 25% have even bothered to jump to superfast speeds. That doesn't suggest there's an ultrafast market anywhere close to that size.
I could be persuaded that the ultrafast market is as big as 10%, I think.
F) Driven by Competition?
There is certainly a prospect for VM to dominate speeds - through Docsis 3.0 and 3.1, but only if VM become serious with their (hidden) network upgrades to increase the number of channels shared through segments. @Ignition believes they are indeed serious. Past behaviour leaves many doubters.
That might be enough to say that BT will have to respond - and concentrate on defending market share in VM areas.
On the other hand, the commercial FTTC rollout seems to have covered just as much non-VM area as VM area, and didn't bother with a significant chunk of VM coverage (more than one-fifth). Maybe they're not too worried.
Of course, VM are expanding, so there will be another chunk of territory to defend. I could see BT targetting places that VM are trying to expand into, for sure.
G) Income Again
There's likely more incremental income (ie an extra £10 per month) from people further away from FTTC cabs
There's likely more old custom to retain by focussing on "project lightning" areas, to prevent loss of all income (~ £40pm)
There's likely some new custom to be gained by focussing on old VM areas.
H) Forget "up to". What about the minimum speeds?
In the FTTC rollout, there have been few anomolies where ADSL gives higher speeds than VDSL2. You'd imagine that, with G.Fast, they wouldn't want too many anomolies showing there either.
If you deploy to the PCP, there is little point in offering service to actual speeds below 100Mbps. You might argue a higher level to that too, for marketing reasons.
That limits the range you would ever want to allow G.Fast deployments. 400m? 500m maximum?
I) Phased Deployments
BT's current plans allow for commercial deployment to run for 8 years - twice as long as the FTTC deployment. That gives them plenty of time to deploy with a phased approach - especially as the hardware continues to be developed in phases.
It might well be true that the initial phases make use of G.Fast's abilities at 300-400m, but future phases might then target higher speeds at shorter range.
How do you achieve that shorter range, while still using the nodes built previously? By placing new nodes at (say) the halfway point, and reducing the maximum range for everyone.
I could therefore see a plan that chooses to deploy DPU nodes at PCP and at 500m. Offering 100-300 with "amendment 2" technology, 200-500 with "amendment 3" technology. Much later, infill could occur at 250m and 750m, turning everyone's lines into something capable of 500-1000.
J) The cost of FTTP
The longevity of G.Fast, and future technical advances into multi-gigabit territory, all depends on the relative cost of FTTP - itself a moving target.
You would imagine that whatever steps are taken with G.Fast deployment, it will always include an element that makes future deployment of FTTP (either native or FoD) cheaper.
There is one line of thought that, eventually, G.Fast will kill itself off by making FTTP cheap enough to be guaranteed to be the next step. There is also a line of thought that, for brownfield, there will never be a reason to force copper out.
I don't know where the end-game will be any more.
K) Overall Thoughts
I think it is far too simplistic to say that "PCP-only" is the deployment style. It has the possibility to be cheapest, but it won't bring in the best income.
If BT are going to attempt "deeper fibre", then the best places to try it are a) Project Lightning areas, and b) VM areas, but they might need to develop their TV offering to make worthwhile inroads.
I don't automatically preclude BDUK areas, as I don't think they will cost more to upgrade (especially in a PCP-only model).
I think G.Fast has longevity beyond the next 8 years, so I think BT will deploy /
this time around/ with a view to what they might do /
next time around/. I can see them having rough plans to deploy at 0m (ie at PCP), at 400m and at 800m. In the distant future, they can infill at 200m, 600m, 1km.
But between now and 2020, I'd say a fair degree of PCP-only (including some BDUK locations), with 400m deployments in VM areas.