One of the areas that policy such as net neutrality will cause some pressure is where limited content pathways can be opened up to deep edges, offering high SLA - for example, an operator CDN model or a multicast network - but (because the capacity of the specific service is limited) it may only become available to specific customers. Within end-to-end IP-only architectures that type of service layout may open up debates around net neutrality.
Many operators are realizing that by using hybrid models, they can, if not bypass, at least compound the complications around arguing that their service is not net neutral. To see this, say a DTT signal is acquired off a traditional broadcast transmission mux remotely at various data center edges and delivered into the IP infrastructure at the edge, without having to traverse the same IP routes as other video signals available at that edge. Then, because the broadcast network is not broadcasting in IP, it cannot be argued that the service was conflicting with net neutrality, even if the DTT broadcast signal was delivered to the data center with a completely different SLA model to the all IP broadcast.
By dropping out of IP into a lower layer-2 telecoms network protocol, it is quite possible that any net neutrality regulations that are introduced may just be worked around by the engineering teams and leave the publisher and end user with unequal service delivery models without actually affecting net neutrality...
... only time will tell. But expect eMBMS / LTE-broadcast services to be scrutinized for this exact situation as soon as the first commercial ones become available.