96 ports promised and even more sought. Burri of Swisscom confirmed he has a similar unit. Update, 7/10. Huawei has now confirmed they are shipping 96 ports. I had guessed it was Huawei in the original. end update. G.fast was designed for local distribution points serving up to 16 homes, of which British Telecom has about 4M. These are very close to the subscriber. G.fast from local distribution points delivers 500-800 megabits today and well over a gigabit very soon. The BT plans changed, going to the (much fewer) existing cabinets.
Many buildings have more than 16 apartments. Many cabinets have hundreds.
Sckipio currently maxes out at 24 ports; Broadcom at 16. Since you can't (yet) have two DSLAMs sharing the same cable binder, that's a severe bottleneck. My building, for example, has about 60 lines on one box in the basement. In Jennie's building, 16 ports would require a vectored DSLAM on all 20 floors, an expensive job.
Huawei, Adtran, and I believe Nokia decided to solve that problem by moving vectoring from the board in the DSLAM to an external box. These were promised for 2016 but the schedule has slipped. I don't know what will be required to move from successful lab tests to volume deployment, but Linney is optimistic.
The large vectoring port count became crucial to BT when the finance people vetoed the distribution point plan. Linney's team did the research that proved 300 meter G.fast was practical. It's now part of the standard.
Long reach G.fast of course is slower, more like 300 megabits at 300 meters. Trevor provided initial data on 340 lines. They showed a wide range of loop losses. The majority of lines achieved the product limit of either 330Mbps or 160Mbps downstream. 99.7% (all but 1) achieved above 100Mbps. BT Openreach is working actively to develop the system for the retail ISPs (including BT Retail.)
They are geared up for the target of 10M lines.