kyrios As you note, Windows is a gateway product, a vehicle to sell other Microsoft products like Microsoft 365, Edge, OneDrive, Azure and so on.
Windows is already on a recurrent fees model, although indirectly, since the largest (by far) share of the Windows market is either (1) OEM or (2) Enterprise/Business/Education licensing. The average lifespan of a Windows desktop is 4-6 years, and a Windows laptop is 3-5 years. Enterprise/Business/Education licensing is typically based on seats, paid/adjusted annually. Both market segments produce a steady income stream, albeit at a significant discount compared to retail, which is not a significant share of the Windows market.
Both income streams, as you also note indirectly, are dependent on third party decision-making, and, in each case, the decision-making process is dependent on Windows having a dominant market share. If Windows did not dominate the desktop/laptop OS market, OEM's would not necessarily deploy Windows as the in-built OS on new desktops and laptops. If Windows <em>products</em> and Windows-dependent third-party software were not dominant in the consumer and business market segments, neither consumer nor Enterprise/Business/Education markets would necessarily use Windows.
The Windows market is fragile, dependent on a chicken/egg relationship between Microsoft, on the one hand, and OEM's and software providers, on the others. It is because Windows enjoys a dominant market share that Windows users at all levels have such a wide choice of software and collateral devices, but it is also because developers provide such a wide choice of software and collateral devices that Windows is able to maintain a dominant market share.
A case in point illustrating both the co-dependent relationship and the fragility of Windows is the Education market. As recently as a few years ago, Windows laptops and Chromebooks shared the Education market on more-or-less equal terms. That has now changed. Chromebooks have secured a dominant share of the Education market (primarily because of simplicity, reliability and lower cost (both purchase and administration), so dominant that even Dell (which provided a full range of Windows-based Latitude 11-3180 laptops for the market) now focuses on Chromebooks for that market. According to industry media reports, the Windows market share dropped roughly 5% in the wake of COVID, in large part because school systems and parents purchased Chromebooks (rather than Windows laptops) for use in remote schooling. The Education market segment is more-or-less lost to Microsoft at this point.
As for consumers, consumers expect a "free" OS (Android, MacOS/iOS, Windows) to come with every device that they purchase, and I expect that Microsoft (and Google and Apple, for that matter) would face overwhelming consumer resistance to buying devices with a "first year free" OS. Whatever Microsoft, Google or Apple do going forward, you can bet that OEM's are going to be resistant to endangering device sales.
I can't predict the future of Windows, but I will be very surprised if Windows moves to a full subscription model in the next 5-10 years, because doing so would cause significant disruption in the OEM and Enterprise/Business/Education licensing markets. I do think that we will see Windows become more of an OS than the catch-all it has become over the years, with the OS segment remaining as is, more-or-less, and "features" moving to a subscription model.
In any event, we seem to be somewhat on the same page in our thinking (although expressing it very differently). Neither of us seems to think that Windows is going to move to a subscription model in the near future, and both of us seem to think that Windows will, at least in part, become part of the subscription model. The important thing is that neither of us knows, but are engaged in speculation at this point.