The main thrust of the article was about monetising software, with the primary question being how can it be achieved? However, I strongly believe that this concern is one that will prove to be a transient issue and will ultimately become irrelevant in the not to distant future.
This will not be because of open-source factors, or because splitting the hardware and the software into two entirely different businesses means that neither business can exist without that fragile symbiotic relationship with the other, but because the fundamental idea of software being produced by people is flawed.
In the not to distant future (and it could actually be done now) all software will be produced by AIs. These AI agents wouldn't need to be as 'clever' as IBM's Watson or need to simulate sentience but would start by just have the ability to correlate simple and well established sets of rules and logic to derive not only better versions of themselves but also applications and hardware drivers. However, as they're doing this they'll also end up deriving new sets of rules and logic which will be shared amongst each other and fed back into this self-development loop.
I would argue that monetisation of such a system cannot work without limiting its functionality, which is fundamentally at odds with how it must work.
Yes, it does sound a bit like the Skynet system in the Terminator films, but supposing that it will inevitably spontaneously develop a human-hating sentience and then try to destroy all the wet-ware is no more rational than believing in magic.
In any case, it's inevitable.