Are we anywhere near that? GeForce Now is solid and xcloud surprisingly has excellent latency (for me), but it’s still there. FPS’s are off the table, not even in a snobby way. You can feel the delay it’s just too severe. The only shooters I’ve ever ever been able to successfully do are slower third person ones and even that’s dicey unless they have other mechanics you can lean on (Mass effect for instance).
I think people get snobby about frame rates but latency is a real thing and there is a threshold you have to cross or everyone notices it, and I’m not sure anyone’s capable of crossing it
> ... latency is a real thing ...
When enough players don't know any different the latency won't be an issue. It'll be the norm, just like using the molasses-slow UI on today's PC is considered normal.
Play a real Atari 2600 on a CRT TV. You'll be amazed at the fluid responsiveness of the system. Use an Apple II or a Commodore 64 and you'll be amazed at the lack of input latency.
We've gotten used to crap latency on our PCs. Gamers will get used to it in games too-- especially if it obviates a four-figure hardware purchase and paying for the privilege of "buying" games you don't own and can't keep anyway.
It's obvious you've never played any shooter game. It can't be the norm, because the game becomes completely fucking unplayable with even 50-100ms of input lag.
Total input lag (a time between a button press and an animation showing on your screen) isn’t just 1000ms/fps, although they do correlate.
Outside of the competitive FPS scene, most people using a gamepad can do just fine with extra 50ms of latency. Some genres hold better, some are worse (mouse lag in RTS games feels pretty shit, but no one plays these anymore).
People are adaptable
Current-gent console market (ps6+switch2+xbox series) is about what, 200-250 million units?
That sounds like a lot, but compare this to smartphones. Apple alone sold 247 million iPhones last year. With, say, 3 year average upgrade cadence it's 600-700 modern devices all the time. Add the high-end models of Samsung and other Android vendors. This is still on-device gaming, much less demanding, where people are fine with worse graphics, lower framerate and uneven framepacing, let alone horrible controls most of the time.
And then add even more people with weaker phones, TVs, slow laptops and old-gen gaming consoles. This is the market I am talking about, and I bet for most of these users a bit of an extra lag is completely okay.
Hardware enthusiasts who are horrifies with GTA6's 30fps is a small bubble- most people will happily play it even on the series S, with drops to 25fps (and an input lag higher than most games deliver on GeForce Now)
Mario is even better as, again, things that are happening in the game are predictable, they just happen tens of milliseconds later than locally. If a game delivers 60-120fps, then playing it in the cloud could be comparable to playing the same game at 30-60fps, and it is FINE. Not amazing- just fine.
What's not fine is the price of the gaming hardware, and some people won't have a choice but to play in the cloud.
Also remember how in the 90s it was considered impossible to create a successful FPS on a console due to sticks being a bad input device compared to mouse. And indeed it was a bad input device for Quake 3, but then came Halo which was built with the gamepad in mind, with aim-assist and a large change in a game design, and it turned out that you CAN make an FPS for a console, it's just going to be a different FPS.
Maybe if we go into the cloud direction we will see a similar change in a game design?
The latency isn’t just a little extra, the frame pacing isn’t even on any service I’ve used (3), and some packet loss is common throughout. So no, it doesn’t work.
Then, earlier this year, tried GeForce now and fininshed CP 2077 Phantom Liberty just to understand the current state of the tech. It was really good.
You started with Mario, which is very playable with a huge lag, but are now switching to Celeste that is indeed an entirely different thing, and I wouldn't want to play it in the cloud, same as N++. The thing is, these games run on a toaster, and most devices suitable to be cloud gaming thin clients can actually function as hybrids.
Steam Deck will run these ultra-fast platformers well enough locally, and for anything more demanding cloud could be the way to go (it isn't right now due to how shitty GeForce now is regarding my own games even within steam library, requiring publishers to grant approval for a game to be available).
For me personally, the main drawback is the lack of ownership and control- I can't play what I already own, I can't install any mods, the game that I bought can be modified without my approval at any time. Probably that's due to my genre preferences, so obviously for you the situation may be different
It doesn’t have to be Celeste. Most Mario games would be very painful with shifting FPS, resolution, and latency.
Cyberpunk is definitely a borderline case. If you have good enough proximity to their servers and have everything hardline/perfectly set up, it is very much playable. But I could tell you that for a lot of people it would not be for all sorts of reasons. Cyberpunk is not a high precision/twitchy game at all. CoD would be unplayable for instance.
Latency is absolutely the number one barrier right now with cloud gaming. The moment a game exceeds 50-75ms or so of latency people start dropping like flies if timing is important (lots of games)
Latency is like audio when it comes to video games. If what you are doing with your hands does not feel reflected on the screen, you will become frustrated even if you can’t articulate what exactly is happening
Mediocre tv speakers are still clear enough. Echo-y/distant/otherwise unclear audio is exhausting and annoying. Hell remember Nolan’s Bane in TDKR? It’s literally all anyone could talk about. People couldn’t understand what he was saying and it was no small matter. To this day Nolan‘s audio mixes are basically a meme that all sorts of people are aware of.
Shrinking demand makes prices go down, not up.
Demand for gaming and consumer gear is higher than ever. Gaming GPUs were expensive before the AI boom. Remember when everyone was shocked that the 4090 was $1500 or the 5090 was $2000, but they were sold out anyway? Remember the extreme competition for 30 series GPUs and scalping before that?
I have yet to hear a good explanation for these beliefs that consumer PCs and gaming are going away, because they’re all driven by incorrect economic ideas (like shrinking demand raising prices) or fear imported from some other ideas, while ignoring the ever expanding market for gaming and consumer PC gear
In a simple supply/demand scenario, yes, but when shrinking demand causes manufacturers to spread costs across a smaller market it can lead to higher prices. A good example is hard disk drives. As the market moved to SSDs the constant cost decreases that had been reliable for decades started stabilizing or even rising. I paid $15/TB for 8TB drives in 2017 and that continued to be a good price for new drives until the AI boom drove costs way up recently. Similar trends have happened with lots of other products as they moved from mainstream to niche.
So prices were low when demand was low, then they increased when demand increased?
If there isn't enough demand for something for multiple companies to find it worth building a factory and producing at industrial scale (and vendors to bother stocking, etc.), then you're faced with either monopoly pricing from the single vendor who sticks around or the dramatically higher cost of other manufacturing methods.
Unlike the B2B market, consumer stuff is usually more stable and history has shown that companies supporting the consumer marked usually last longer.
The given the demand things could move as quickly as the AI psychosis inflated the bubble, hopefully, to satisfy the demand.
Looking forward to cheap AI chips and HBM on custom board from scrapped AI datacenters - with next day shipping from Aliexpress warehouse in the EU. :)
Non-gamers will try pitching this over and over again, without ever learning from their past mistakes.
Streaming doesn't work for gaming. Any gamer can easily feel even just 50ms of latency in the form of input lag, and it makes the controls feel terrible. Gamers won't accept worse game feel -- they want better games, not worse ones.
How many countless game streaming platforms have to fail before you guys realize it will never work?
(Don't get me wrong, I'm not excited at the prospect of being forced to stream games, but i don't think technical infeasibility will protect us for that long)
I've been hearing this for over two decades and yet gaming desktops are more mainstream than ever. Certainly the current market is negatively impacting pricing and availability, but those same factors are also pushing their competition just as hard. I'm not even sure cloud streamed games was a profitable idea when prices were sane, but it certainly isn't now.
On the business side I've got Customers looking to replace servers purchased in 2020 and 2021 and they're not able to get anything reasonably similar for less than 4X what they paid for the old machines.
Components are still available to buy, take a look at any marketplace. Yes, components are rather expensive these days, especially RAM, but perhaps I should remind you that in the '90s I paid something like £30 per megabyte. Then again I can now buy a ESP32S3 for a couple of quid.
Anyway, its horses for courses. With a small amount of imagination you do not have to rent anything.
You are completely wrong about storage costs which are now nearly negligible compared to RAM.
I still have my Commodore 64, which now sports a USB interface. I also play games on my phone and that is still working OK, without a sub. My Clash of Clans Town Hall is level 17.
How much more "enthusiast" do you need from anyone who really couldn't give a shit about gaming?
That's not what I'm seeing in the pricing trends: https://pcpartpicker.com/trends/price/internal-hard-drive/
Every tracked storage device there has trended up dramatically over the last year. I see the same trends on other price-tracking sites, too.
I recently bought a trio of reconditioned Supermicro boxes with enough flash to do a Ceph cluster of 80TB with 2/3 redundancy ie "RAID 5" as well as run the rest of Proxmox. I had to dial back the RAM to a mere 128GB each node to get the cost under £15,000. I also slapped a double QSFP28 NIC in each (100GB/s) and a pair of 10GB copper NICs.
Storage, CPU, and network are relatively cheap compared to RAM.
That's entirely meaningless unless RAM requirements also go back to 90s levels, which they won't. The unit of RAM people are buying isn't measured in MB but in 'can it run what I want to run'.
There might be some unexploited latency-masking tricks in client-side cursors and client-side reprojection (the timewarp trick some VR uses to rotate the camera without receiving a new frame), but I suspect the legacy internet connectivity of rural and suburban homes in america will be a serious barrier to widespread adoption of real-time streaming. If local hardware gets priced out of reach, it could rather mean the end of gaming as a widely accessible hobby instead of a switch to real-time streaming?
It's both "local" and "in the cloud" if it's a subscription to a browser-based game.
Genuinely asking I’m not sure of the state of the art here. I tried out GeForce now a couple years ago and it was fine for couch co-op platformers but I could never imagine using such a delivery method for an online FPS game.
I think the bottleneck there is consumer adoption of gigabit fiber and data center distribution. If i had to stream from a west coast server and not a Midwest server there's nothing that could bring the latency within 1 frame no matter what.
> Gaming will move to real-time streaming
the main problem is not existence of cheaper options - but lack of advertizement
there are 20$ gpus, ffs, nobody is simply aware of them
Gamers can’t even tolerate bluetooth/radio lag which is why they use cables for their mouses/keyboards.
Like I have both Stem Deck and Steam Machine & they work so nice for gaming, have robust OS update mechanims, yet still don't limit me to use it for any tinkering, development or running of non-Steam games I want to do.
I have no idea why anyone would ever buy a locked down PoS console with overpriced games in 2026.
I would say we already have been doing that with Web 2.0 and it's going to trend the opposite
Been building computers since 15, i'm now 40.
Unless they take every physical component off the shelf and prevent it's sale, how can it die?
Then there is the second hand server market. Will old servers not be sold anymore?
As long as there is one shop selling components (or one person on a marketplace), I can build computers. ill use DDR3 if I have too.
What remains? A SBC board + a case? Is that still called "buiding a computer"?
Home computing, in the context of having a marginally-powerful computer or phone / tablet with minimal storage and RAM in the home, will continue just fine. Subscriptions for storage and applications are already becoming the norm, and with the skyrocketing cost of storage the market for large capacity storage targeted at enthusiast and home users will dry up, too.
The industry wants users to rent everything.
Storage rental is particular pernicious, to me, because being able to store your personally-created works in a manner that maintains confidentiality is to cornerstone of digital sovereignty. Being forced to store your data in systems that will use your data for model training is definitely desired by the industry. If they can con users into renting that storage and giving up their data to model training so much the better.