I got the point of the article, but I kind of agree with @JesperMP that the analogy fits only very roughly.
What are the similar advantages that soft-plcs brings to the market that cannot be done with traditional PLCs ?
The article mentioned some tools and techniques that the soft-PLCs bring to the table: containerization/Docker; VMs; Git; Object-Oriented Programming (OOP). These will matter to developers, and to high-end users controlling complex tasks; for a process that can get by with a set-it-and-forget-it ladder running a Micro8xx or S7 1200 with minimal, or no, HMI, they would be a waste of money.
If these "advantages" are indeed aimed at a particular segments of the market: developers and high-end users, then this does not fit the "Innovator's dilemma" model, where the new tech encroaches initially on the low-end of the market. And while that may be enough to grow market share, I doubt it is enough to be violently disruptive (yet).
Git, and Gitflow in particular, are disruptive, but for
developers. Git might be useful at the plant level but mainly as a way to restore functionality if something breaks down; however development and recovery are capital expenses, so this is not where the money is made.
I'm not sure I see this as much of an advantage; what containers bring to the table is a way to configure an environment inside an OS that does not involve configuring the host OS, and isolating the environment from the host OS. So if driver A would incompatible with driver B in the OS, put them in different containers or VMs. Also, they provide a way to put all the configuration in one place: if a soft-PLC breaks down, then I can replace the hardware and re-provision the containers, without as much concern for the firmware revision number. Being able to quickly replace failed hardware can be useful when downtime is measured int k$/h, but a decent spares and bakcup plan with hard-PLCs makes this less than disruptive.
I like programming, but more or less dislike all programming languages.
That said, each has its application E.g. ladder is great for the customer who outsources development but handles maintenance and troubleshooting internally: a tech is going to be able to troubleshoot a ladder diagram of contacts and coils, but will be lost if given CodeSys (or Python).
If a process can benefit from OOP, then it is likely a high-end process. CodeSys is not exactly elegant, to put it politely. If I had to do something complex on a soft-PLC I would prefer Python or C/C++, where, clumsy as they are, at least there is a plethora of libraries available to solve complex problems. But again, that is from a developer, not a low-end user with a simple process to run.
One thing this might do is push business logic closer to the process; with hard-PLCs that usually requires the additional cost of SCADA-like or edge solutions.
Another piece of this is reliability. There is something to be said for using a hard-PLC that has limited functionality with correspondingly limited ways it can go wrong, and has been tested with a large corporation backing it up. Am I going to call Linus Torvalds and say "Your OS put a couple of lift forks through the side of my tank. Whom do I sue?" And then later to find my own memory leak that caused it?
One advantage not mentioned is the possibility of full-featured web-based management. Every web-server on a hard-PLC I have seen is clunky at best, and tied to obsolete tech like ActiveX controls. To be able to manage, to query, to use a web-API, to build custome web interfaces and HMIs, to even program with any web client instead of expensive and version-sensitive applications, with real security, might be a significant benefit; if anything, this is the killer app.
I made the case above that the soft-PLCs are not aimed at the low-end of the market. That is not strictly true: something like a RaspberryPI-based soft-PLC might be a cost-effective capital alternative to a low-end hard-PLC brick, but that savings would be a negligible part of any commercial implementation. So I think that market is mainly for hobbyists who can with a clear conscience put a low value on their time.
Soft-PLCs are a growing part of the market, but I am not sure they are disruptive according to the "Innovator's Dilemma" model. The possibilities of the all-in-one, web-managed PLC+HMI might get it there.