GF's marketing for 12nm gives 84nm for contacted gate pitch and 64nm for metal pitch, same as for its 14nm LPP.
https://blog.globalfoundries.com/gfs-12lp-process-behind-covers/
Samsung's marketing for its 14nm LPP has it at 78nm CPP, which is a notable difference.
https://en.wikichip.org/wiki/14_nm_lithography_process
However, AMD at one point gave Zen's CPP as 78nm.
https://hexus.net/tech/news/cpu/102274-amd-ryzens-x86-core-10pc-compact-intels/
Samsung's 11nm differs from GF's 12nm in that the former chose to use the 10nm node's metal pitches (while also having an option for CPP at 84nm like GF).
So there's kind of an odd mix to the marketing of the respective processes, where there's potentially some room to more closely match but not with what they give as their public offerings.
As for what AMD gains, it is curious. One possible interpretation is that if AMD expects its 14nm CPUs or Rome's 14nm IO die to eat into allocations, nudging Polaris to 12nm and Samsung may give more flexibility.
I don't see any projections that Polaris is going to set the world on fire at this stage, so perhaps it can hold the fort and fulfill some amount of WSA capacity in this scenario until it is replaced.
There could be a pipe-cleaning use for second-sourcing to Samsung, or perhaps this is throwing a bone to Samsung, since AMD mentioned working with them as a possible second source for Zen during its ramp.
Also, this can serve as leverage for WSA negotiations. AMD's position is likely stronger if it can meet certain product obligations without GF.
