Basically, the Gates view was the more "correct" one for the 1980s/1990s. At that time, most people did not own personal computers, even many businesses didn't, and growth came from on-boarding these non-users. The more apps available on your platform, the more likely a non-user would be to see a reason to buy a computer at all. Also, the platform Gates was selling was much cheaper to try.
Today, everyone owns at least one computing device. Platforms aren't competing for non-users (there are none left), they are competing for users of other platforms. The Jobs view works better in that world since it improves platform quality perceptions.
Saas is the computing equivalent of an airline, and it's very popular, but it hasn't fully displaced individual computer ownership. We'll see what happens long term.
The writing is on the wall though, isn't it? Even companies that you'd expect to go all-in on user freedom and ownership have locked down their platforms and ratcheted up the service offerings. Not a single FAANG company is willing to take a definitive stand in defense of their users. They all just see a bottom line waiting to be butchered.
And for most users, I'd wager SaaS has displaced computer ownership. Stuff like the App Store, Play Store and even game consoles are ultimately a service unto themselves. Even disregarding that, the "happy path" of modern software is paved far away from user empowerment. People are generally more willing to pay for a subscription than to switch to a more convenient hardware platform.
I'm really not sure. Five years ago, I thought that SaaS would completely replace local computing. Now, I'm less certain.
Yes, SaaS is very popular and there are far fewer local applications now. However, there still seems to be some value in local applications because it's thought to be a better end user experience. Look at Slack. It started out as a website, pure browser-based SaaS. Over time they developed a local app, which is basically the same Javascript but packaged up as a local Electron app. This seems to be regarded as superior to the web version for most people.
Consider Zoom. There was initially a native client and an equivalent web version. They seem to be actually killing off the web version now -- access to it is a non-default option for the meeting creator, buried on a screen with 100+ meeting settings nobody understands or reads. They apparently want to be a local-only program.
As long as there is demand for local apps, people will continue owning general purpose, re-programmable microcomputers. Maybe the vendor can lock you out of some low-level stuff like Intel ME or the BIOS, but these platforms are still allowing people to download a compiler and run their own programs directly on the metal.
I'm not sure what direction it will ultimately go, but my hunch is that if it were possible for SaaS to 100% displace local computing, it would have already happened.
To counter that though, I think we have to define SaaS concretely and ask what a more service-driven world would look like. I consider SaaS to not only be the paid components of an OS, but also the "free" services like Google and the App Store. Holistically speaking, our OSes are saturated with services; Find My, Android Notify services, OCSP/Microsoft/Google telemetry, online certificate signing, and even 'assistants' that fail to assist without internet. What more could we possibly put online?
It's a bit of a rhetorical conclusion to leave things on. I'll certainly see worse examples in my lifetime, but the status quo is bad enough in my opinion.
Anyway, sure, let's leave it there and see where things are in 5 years. It'll be interesting to find out!
If your need was niche then go away.