They don't intend to be left behind. AT&T's Tom Starr is one of the key engineers behind G.fast, which was developed based on the plans of British Telecom and AT&T. Their top execs have been enthusiastic about G.fast but carefully didn't commit. January 4, 2017, they announced "Experiments expanding" and that they were "encouraged" by results in Minneapolis. Cable is making strong inroads against U-Verse and destroying AT&T where they never upgraded decade-old DSL. They have to do something. They intend to abandon landlines and go completely wireless to 10M homes. The 12.5M fiber home lines are only about a quarter of their future footprint.
G.fast from basements and rooftops is the natural cost-effective choice and has strong advocates within the company. AT&T spent $billions expanding fiber to 1,000,000 business addresses. Much of that is very close to residential buildings and easy to extend. AT&T is speeding up their fiber home build, now at 4M. The 12.5M target is now moved up to mid 2019. They acquired millions of lines connected by coax to a DirecTV rooftop antenna. G.fast performance over coax is exceptional.
I've reported that T is actively looking for more buildings for WTTR, which likely will migrate from dedicated lines to G.fast.