> What I don't get is, what's in this for Sprint and T-Mobile?
Both Sprint and T-Mobile have to build physical infrasture, both to offer basic services (e.g. 3G, talk/text) but to also compete (4G, LTE, etc). Building infrastructure is really expensive. They need land, licences, power, and a network to connect that tower to (assuming it isn't just a relay).
So both networks have this HUGE sunk cost. In an ideal world they would be able to recruit enough users to consume each tower's capacity by 100% at peak. But because they're competing against other networks (with more users) and have to look competitive, they've over-built their network both in scope (e.g. servicing areas with few users) and capacity.
If Sprint and T-Mobile were confident that they could recruit enough users to fill their network they likely wouldn't agree to this deal. Just as AT&T and Verizon haven't. However the likely reality is that both Sprint and T-Mobile are well over capacity in certain areas and every penny that Google can bring in by stealing uses from the big two is just further helps Sprint and T-mobile, even if they just use it to further build out their respective networks.
Additionally: You could frame this as an "illusion of choice." You have Walmart/Target Cellular (Straight Talk/Brightspot) on the low end, T-Mobile themselves in the middle, and Google's Project Fi offering a niche service, but in reality they're all just T-Mobile aimed at a different user base. In all three cases T-Mobile grows, their network grows, and they could kick these brands off any time the contract is up for renew.
PS - T-Mobile actually host just under 30(!) different cellular providers in the US[0]
[0] https://en.wikipedia.org/wiki/List_of_United_States_mobile_v...