But running a SC does have overhead and risk. It seems obvious to me that companies will use the mainchain if possible (that is, it scales and can capture the transaction).
There are significant number of people who don't want the main chain to scale, for various reason.
Sidechains provide those people both with an argument against scaling the main chain, and it creates a group of people with a financial interest in preventing the main chain from adopting new features which might render particular sidechains less necessary.
The latter part wouldn't be such a big deal if Blockstream wasn't so closely connected with Bitcoin Core development.
I think most of the sidechain.pdf authors still want to hard-fork bitcoin to support an increased max block size, no? On that matter, are there any core devs remaining who are
against increasing the max block size?
Nevertheless, I made the same point earlier that sidechains could be used as an "excuse" to not move forward with this hard fork, fragmenting community opinion at a time when we need to come together to support this change. Ideally, I like to see us move forward with the hard fork within the next 12 months, sidechains explored using federated servers, and the risks and rewards of adding support for SPV proofs to the protocol analyzed in much greater detail from the technical perspective, but also from the economic, legal, and game-theory perspectives too.