I have been talking to a number of cable execs myself recently, and my attempt at The Truth is something like this:
The Medium. In general, the shortcomings of HFC networks include limited downstream (where Docsis and other technologies help out), even more limited upstream (where channel bonding options are more limited), the medium itself (which is much less 'transparant' for signals than fiber, which is why HFC operators need several amplifiers in their access networks) and the fact that it is a shared access network (just like wireless).
The Upgrades. HCF operators have a choice of many technologies and techniques (I counted 14!) to upgrade their networks. The trouble is, some do not apply for practical reasons, some are still embryonic and others may prove costly.
End-game. Even if HFC operators manage to do node splits to 1 per 20 homes and fiber deep to reduce the access network to 50 meters, expand the spectrum to 3,000 GHz and apply 256-QAM, questions remain: how much does it cost, and is it enough?
Access network. In the above case, a 50 meter access network could in practice not be too different from an in-home network based on copper or coax in most FTTH networks.
Non-linear video. The big threat is a migration away from linear TV to non-linear HD video, in both the uplink and the downlink. Here it is important to note that people tend to overestimate the short-term and underestimate the long-term. In other words: yes, upgrading will allow HFC operators to compete for several years; and no, it may not be enough and a full FTTH migration may be necessary.