The question here is why you would host such a device for free instead of just having SFI peering with apple (which is apparently required in this program). I suspect that the incentive there has mostly nothing to do with transit costs and is mainly about capacity planing inside the ISP's network and thus with fixed hardware and infrastructure costs. (ie. SFI is free, but the 100Gbps port on your router is not)
It may be possible to place caches deeper inside an ISP's network than the peering points. For example, it looks like Apple peers in Dallas but not Austin or Houston so putting a cache in Austin would save bandwidth up to Dallas.
In my experience it will invariably be placed deeper into the network, that is at least into network core of the ISP, which typically isn't anywhere near the edge router placed in some wonderfully expensive colo space associated with some IXP.
Completely right, it would've been more correct to say "saves money" and not specifically call out transit.
Edit: the flip side of this relationship is also interesting - if Apple or whoever doesn't offload enough of their traffic to the rack, then it isn't cost effective and can really annoy the ISP. I've known some ISPs to boot these caches out of their network when the related company wasn't utilizing it effectively.