>thin clients that rely on an internet connection to be used
It's already a thing, for programmers the model of just spinning up a remote server instance and connecting to that has gained a lot of popularity (see e.g. github codespaces which brings that model to the masses). This also coincides nicely with a move towards more power efficient processors (move to arm), where you can just offload anything really compute-heavy to a remote server while still maintaining good enough local performance for most tasks, at a fraction of the power draw.

> I can think of which would prevent thin clients
Latency is only an issue if you're using a protocol that relies on streaming back graphics. With things like VSCode Remote, the typing latency is not an issue because the client does in fact maintain state, and syncs it with the server as needed. There is however still some latency in things like fetching autocompletions. (In case I'm not clear, think of the difference between mosh and ssh with regard to state maintained by client). Even JetBrains (famous for their monolithic and feature-packed IDEs) is moving towards facilitating thin-client setups.

> There's also the issue of gaming
There used to be Stadia, and from reports the latency was tolerable for the vast majority of people. I don't expect people playing competitively to use it, but technology-wise it seems to be mostly ready. Probably the bigger issue compared to a technological one is a trust issue, there's a sizeable fraction who have a distrust of not actually "owning" the game, and ironically Stadia just reinforced that when they shut down.


I think AMD would be fine, they're not really in the same domain as companies like Noctua/Crucial.

I think there'll be more of a market for PC handhelds like Steam Deck, and custom built PCs will be sceen as a niche for enthusiasts (similar to audiophiles). Even content creators nowadays can do just fine with a laptop or even tablet. Console gaming might go out of style unless companies go back to creating more toy-like consoles like the Wii. People seemed to like that.

Also, I have a feeling AR will take off soon, just not sure when. Lots of apps with a focus on photography have filters that add stickers onto the face for example, hell you even see it in some mobile games, so I thought by now there'd be more companies trying to one up another over AR.
Guess we will see whether Apple can innovate AR or not soon.

>>3231
"Thin clients" have always been a thing in the workplace, workers just use whatever they can in the office to connect to another server/service and work from there. The only thing that's different now is that they got much smaller.

>>3232
>content creators nowadays can do just fine with a laptop or even tablet
Docked laptops may get more popular.

>Apple can innovate AR or not soon
I'm guessing you mean VR instead of AR? AR can be done without the immersive VR headset, but there haven't really been any killer apps for that besides overlaying things onto faces or measuring. VR+AR probably will enable interesting interaction paradigms, and at the very least if the resolution is good enouogh it should be able to replace a monitor entirely. My only concern is from a health perspective, I don't know what the long-term effects of using VR would be. The naive argument would be to dismiss concerns of myopia given that they use fancy lenses to make the focal length longer than it actually is, but the mechanism by which the eye initially grows is not fully understood. If it was purely focal length then the reduced lens method would work for myopia, but a crucial part of it seems to be expose to sunlight. So the eye might actually be looking at some difference in the aberration of wavelengths or something else.

>>3231
I wouldn't call use by programmers and within a corporate setting mainstream. Seems more like something they have to use more than want to. I know someone who uses VMWare at his job for security reasons, and it's so laggy and disconnect-prone, he barely tolerates it.
>>3232
>PC handhelds
I don't think this is a real thing, just a marketing term. How is a PC handheld fundamentally different from a regular handheld? The library overlapping with PC doesn't matter if PCs go the way of the dodo. Lots of console games get ported to PC, but you wouldn't call your PC a "desktop console".
>Console gaming might go out of style... toy-like consoles like the Wii
The Switch actually outsold the Wii by a large margin, and it also incorporates a handheld form factor. It even outsold the original Game Boy. So I don't think "conventional" consoles are in any danger.

>>3234
The VMWare "stream your screen" paradigm is quite out of date as I mentioned, modern (e.g. LSP based) client-server setups incur no additional typing latency.

>>3235
>LSP based
And what about security? The entire purpose of using VMWare is to isolate the work computer from the client.

>>3236
The only communication between client and server occurs through LSP, so there's no security issue here?

>>3237
The point I think he's making is that there's way more more "semantically meaningful" information handled by the client (i.e. text buffers, codebase metadata, etc) with the "LSP" (specialized client + RPC mechanism, really) approach, which might make it easier to leak remote data (either intentionally or, say, by local caching and whatnot), as opposed to the desktop streaming approach, where all you have is a bunch of pixels from your remote screen and all of the meaningful state is on the remote end.

>>3237
At my dad's job, he uses all kinds of software, not just an IDE. Database managers and all kinds of built-in-house stuff. They also use Excel files stored in Windows network folders, and stuff like that. None of that can be accessible from his host OS.

>>3238
Yeah that's fair, although I find it hard to imagine a threat model where an adversary can access in-memory text buffers but wouldn't be able to just capture the entire framebuffer (thereby accessing remote content). Usually the threat model with regard to thin-clients for FAANG companies is to avoid any code hitting disk, which is fairly easy to guarantee even with an LSP-type approach.

A few things have come out in the months since this thread was made. First of all, ARM CPUs are coming to the mainstream. Apple has already made the transition, but now both NVIDIA and AMD are planning on getting in on it. I think this will constitute a new "era" of personal computing, ending the one that started with Windows XP.

On the software-side, compatibility with win32 applications will become less emphasized, and maybe even gradually phased out. Compatibility layers are bound to be imperfect, and most users don't care enough about that for it to affect their purchasing decisions. Web-based applications have become popular even in the corporate sphere, and while there's a few irreplaceable professional-applications, all of those will get ported. On top of all that, emulating a windows 7 machine is becoming increasingly trivial on modern hardware.

On the hardware-side, lower-end GPUs will probably go extinct as iGPUs get better. Power efficiency might become more emphasized, which ARM contributes to. Those 1300W PSUs might disappear. Socketed ram might also go extinct because there's performance and energy advantages to coupling it together with the CPU. Most people upgrade everything all at once anyway. It really depends on how AMD and NVIDIA decide to handle desktop ARM.

The gaming industry as a whole doesn't seem to be doing too well. Games have become too expensive to make, and they're kind of crap now for the most part. I don't know how much interest there is among the general populace in non-mobile stuff. This will affect PCs too, but I'm not sure how and to what extent.