Good news indeed for investors. The AI craze is upon us and Nvidia is at the helm. Nvidia stock has soared incredibly over the last 5 years by 2,500% and their Market Cap is now 2.4 trillion USD making them the 3rd most valuable corporation in the world with MS at #1 and Apple at #2.

I am interested in where lies my one and only concern with Nvidia. How much will they care about the gaming market now? Will they even be concerned very much with entry level through midrange gaming customers considering the relatively small profit margins and the available wafer space at TSMC? Then the pricing. Will they jack up the prices yet again like they did with Ada? We will get some idea before the end of the year but it will be next year before the entry level and most of the midrange arrive to see.

Nvidia Gaming sector y/y is +18% while AMD is -48% (expected to go down even further in H2 due to low console sale).

I guess Nvidia could just put every research onto their AI super-computers and call it a day

64KI am interested in where lies my one and only concern with Nvidia. How much will they care about the gaming market now? Will they even be concerned very much with entry level through midrange gaming customers considering the relatively small profit margins and the available wafer space at TSMC? Then the pricing. Will they jack up the prices yet again like they did with Ada? We will get some idea before the end of the year but it will be next year before the entry level and most of the midrange arrive to see.

Nvidia seems to care enough about the gaming market. It is a good steady baseline market while Data Center - read: AI - had a 5x increase YoY. This is a boom, how high it gets and how long it lasts is still unknown and while Nvidia hopes it is a permanent thing I doubt they put all their eggs in that basket. A lot of the R&D carries over across the segments anyway.

TSMC capacity availability is a much more serious problem though. They and everyone else will go for high-margin things and moreso when we are talking about cutting edge nodes which inevitably will leave Gaming a poor step-child.

Entry-level is dead. Has been for a while and is becoming more so. iGPUs are getting faster and APUs have already taken over practically of of the entry-level market.

Gross margin close to 80% is a good measure of how strong of a monopoly Nvidia controls. Anything over 50% is great but 80% is just obscene.

Nvidia might exit the individual discrete GPU gaming market in favor of cloud gaming especially if the AI market corrects a bit and they need another use for all that computing power. SoCs are all the rage now and all you need to connect to server farms thru widely available high speed internet.

DavenGross margin close to 80% is a good measure of how strong of a monopoly Nvidia controls. Anything over 50% is great but 80% is just obscene.

Nvidia might exit the individual discrete GPU gaming market in favor of cloud gaming especially if the AI market corrects a bit and they need another use for all that computing power. SoCs are all the rage now and all you need to connect to server farms thru widely available high speed internet.

Nah, discrete GPU isn't going away for consumer, and cloud gaming will fail. Its not taking off. It never has and never will due to the dependancy on service quality and general latency, plus the fact you're now subbing yet another service. The moment those cloud services start billing the real cost of the service they will slowly die. Its too costly and any cost reduction will inherently make the service worse.

If anything, we'll see the slow trend of lower tiers being relegated to APUs continue.
So really more choice instead of less.

Vayra86It never has and never will due to the dependancy on service quality and general latency,

I definitely see a fork coming between casual gaming and esports. Latency is a big part of that of course. I remember the glory days of Yahoo games. So many hugely popular browser games.



We’ve come a long way since those days. If approached properly, game streaming can work. Avoiding an all or nothing strategy is key. Nvidia also seems to think so with its Geforce Now service. I’m sure they would rather have the Game section of its earnings report become mostly subscriptions.

londisteTSMC capacity availability is a much more serious problem though. They and everyone else will go for high-margin things and moreso when we are talking about cutting edge nodes which inevitably will leave Gaming a poor step-child.

^^THIS^^

Between AI-everywhere/everything & Apple, who has recently announced their AI plans, the proverbial sh^t is gonna hit the proverbial fan real soon, since there are already significant backlogs in capacity, so yes, entry level gammin is & has been a very poor, wet, wimpering step child for a while now, so you can save your tears for some other sector of the tech industry :D

For now, I still eagerly await my ever-increasingly larger dividend checks, but whenever this bubble busts (as all tech bubbles do eventually), then we will have even more serious sh^t to worry about..

*NOTE* please see my signature for the only real solution to this problem :)

DavenI definitely see a fork coming between casual gaming and esports. Latency is a big part of that of course. I remember the glory days of Yahoo games. So many hugely popular browser games.



We’ve come a long way since those days. If approached properly, game streaming can work. Avoiding an all or nothing strategy is key. Nvidia also seems to think so with its Geforce Now service. I’m sure they would rather have the Game section of its earnings report become mostly subscriptions.

Not just E sports though, you can't fork this properly.

Browser games vs casual gaming is already problematic, they're different things. Is casual gaming also the guy on a console playing shooters? Where is that line drawn? Are businesses going to sell the line 'you're a casual, get it in cloud'? I'll have my popcorn ready for that!

Vayra86Not just E sports though, you can't fork this properly.

Browser games vs casual gaming is already problematic, they're different things. Is casual gaming also the guy on a console playing shooters? Where is that line drawn? Are businesses going to sell the line 'you're a casual, get it in cloud'? I'll have my popcorn ready for that!

How about enthusiast vs professional? I mean most already accept amateur vs professional in the sports world even though amateur has negative connotations. Besides, I don't think we have to label people anyway. If a AAA game plays at 60 fps 4K over streaming at least as far as the observer is concerned, I think most will be fine with that. Not people here of course. TPU readers will never be satisfied. :)

DavenHow about enthusiast vs professional? I mean most already accept amateur vs professional in the sports world even though amateur has negative connotations. Besides, I don't think we have to label people anyway. If a AAA game plays at 60 fps 4K over streaming at least as far as the observer is concerned, I think most will be fine with that. Not people here of course. TPU readers will never be satisfied. :)

If anything happened in the PC market for gaming, and gaming as a whole, it is exactly NOT unification in terms of output, but diversification.

Some people want gaming on the go.
Some want their gaming from the cloud so they don't have to spend on hardware.
Some want everything local.
Some want the lowest possible latency to be competitive (a substantial group, certainly bigger than what's happy on cloud today)
New markets have evolved out of E sports and resulting PC gaming visibility. PC has evolved into a fashion statement and of course they won't buy that hardware to log into a cloud server, you'd look like a fool and that's exactly what this customer group is sensitive to - if you can't show off the stuff locally, what's the point?

Gaming caters to such a wide variety of use cases and demographic now, cloud can only cover a small part of it. And it probably will. But not to the detriment of the rest. Its the 'PC gaming is dead' in a new coat, and we all know that ain't happening. Also look at the consoles. They're still selling the dream of local hardware and locally ran games. MS is trying to move away but is highly cautious, because they're about to give Sony an even stronger USP that way.

And in addition to all of this, the old and new generation is now starting to feel the impact of services suddenly discontinuing. Initiatives rise to counteract that trend, because people now feel what it means to get an investment (of time, effort) ripped out of their hands at any random given moment.

Its a corporate fantasy. Nothing else.
It will line up in the failed projects alongside autonomous driving, the current batch of AI, and VR. And why? All for the same reason: the human factor. We're not computers and we don't walk through spreadsheets or align with strategies. What seems logical in stats, won't be logical in the real wild called life. What I like today in cloud, is likely overruled tomorrow by something that shows me the problems of cloud. I could drive autonomously for years, but when that car runs a single dog over, my world changes. Whereas if I drive the car myself, I'll keep doing so.

The bottom line: if we depend on services and on others, they must do exactly what we expect them to do. And if they don't, we get rid of them, or correct them if able. There is no system that does exactly what we expect it to do dynamically - we expect input = output. We want control.

Vayra86Nah, discrete GPU isn't going away for consumer, and cloud gaming will fail. Its not taking off. It never has and never will due to the dependancy on service quality and general latency, plus the fact you're now subbing yet another service. The moment those cloud services start billing the real cost of the service they will slowly die. Its too costly and any cost reduction will inherently make the service worse.

If anything, we'll see the slow trend of lower tiers being relegated to APUs continue.
So really more choice instead of less.

Yeah discrete GPUs aren't going anywhere.

I really do hope there's a snapback against everything being a subscription nowadays, that's a cancer.

DavenGross margin close to 80% is a good measure of how strong of a monopoly Nvidia controls. Anything over 50% is great but 80% is just obscene.

Yeah their gross margins are, well gross. Apple's highest gross margin was 47.4% (a figure they aren't far off now) and yet they close to HALF Nvidia's gross margin. That's insane. People used to balk at Apple for their margins but Nvidia is in a different league.

evernessinceYeah GPUs aren't going anywhere.

What does that mean? Who said anything about discontinuing GPUs?

We are talking about streaming versus local gaming with respect to Nvidia future product and service business strategies. GPUs are still needed for cloud gaming. Lots and lots and lots and lots and lots of them. This conversation comes from the fact that Nvidia data center is 10x the revenue of their gaming category. This could lead them to change, deprioritize or outright exit current ways that they deliver said products and services.

Being able to go to the store and buy a PCIe card called a GPU is totally based on market directions. If you asked someone ten years ago when the optical disc drive of any variety in any device would disappear, most would say never. Same goes for digital non DSLR cameras and even those are close to extinct.

DavenWhat does that mean? Who said anything about discontinuing GPUs?

We are talking about streaming versus local gaming with respect to Nvidia future product and service business strategies. GPUs are still needed for cloud gaming. Lots and lots and lots and lots and lots of them. This conversation comes from the fact that Nvidia data center is 10x the revenue of their gaming category. This could lead them to change, deprioritize or outright exit current ways that they deliver said products and services.

Being able to go to the store and buy a PCIe card called a GPU is totally based on market directions. If you asked someone ten years ago when the optical disc drive of any variety in any device would disappear, most would say never. Same goes for digital non DSLR cameras and even those are close to extinct.

Sorry, my mind forget to add the discrete part.

We can feel how much attention Nvidia is giving to the consumer market, being ahead of competition and selling well even with high prices...

Some of that gaming revenue are 4090 sales which aren't being used in gaming. So you can't directly compare probably with AMD GPU as they are all for gaming and aren't useful for small LLMs.

More importantly, datacenter isn't an end user category. Whoever is buying them, has to create a product an "AI" product, then sell that on to an end user, be it software using LLM to cut costs, a chat bot or whatever.

I'm interested to see what the use cases are, that will generate $100bns+ a year in revenue, because I don't see them yet myself. Looks more like 3dtv (manufacturers creating a product that no one really wants), on the consumer side anyway than something that people are demanding, like they did with other big consumer breakthroughs, like the iPhone or similar. Then the other issue then is if whoever is buying can ultimately pay Nvidia for the product or if a lot of money that is supposed to be receivable becomes bad debt.

DavenBeing able to go to the store and buy a PCIe card called a GPU is totally based on market directions. If you asked someone ten years ago when the optical disc drive of any variety in any device would disappear, most would say never. Same goes for digital non DSLR cameras and even those are close to extinct.

ehhhh I guess you aren't following the cameras market too much? Basic point and shoot are dying, but the high end/exotic is thriving. Fujifilm can't keep up with the demand for the X100VI, and the older X100V is being scalped. And that's not even their biggest moneymaker: that little camera represents over 50% of Fujifilm imaging business. And they are seeing yearly double-digit growth.

Instax is More Than 50% of Fujifilm’s Imaging Business and is Still Growing | PetaPixel

• Strong sales of instant photo systems and digital cameras boosted revenue by 14.5% year-over-year to JPY469.7 billion and operating income by 39.9% year-over-year to JPY101.9 billion.
• In the Consumer Imaging business, steady sales of the INSTAX instant photo systems drove revenue higher. INSTAX's sales target of JPY150 billion for FY2025 was achieved one year ahead of schedule.

Fujifilm Announces Financial Results for Fiscal Year Ended March 31, 2024 | Fujifilm [France]

There's some market where customers are resisting the unification/digitalisation that many tech companies are dreaming of. Apple made an ad suggesting that the new iPad would replace real instruments, cameras and so on...Apple had to apologize because it wasn't well received: Apple apologizes for ad that crushes the sum total of human artistic endeavor | Ars Technica

People still want to have fun, to interact with something "real"even if it's "less convenient". As long as there's a fun factor, a rush of dopamine, people will seek it. Something that many marketers are failing to realize. And I think that ownership of gaming hardware is part of that.

mb194dcSome of that gaming revenue are 4090 sales which aren't being used in gaming. So you can't directly compare probably with AMD GPU as they are all for gaming and aren't useful for small LLMs.

More importantly, datacenter isn't an end user category. Whoever is buying them, has to create a product an "AI" product, then sell that on to an end user, be it software using LLM to cut costs, a chat bot or whatever.

I'm interested to see what the use cases are, that will generate $100bns+ a year in revenue, because I don't see them yet myself. Looks more like 3dtv (manufacturers creating a product that no one really wants), on the consumer side anyway than something that people are demanding, like they did with other big consumer breakthroughs, like the iPhone or similar. Then the other issue then is if whoever is buying can ultimately pay Nvidia for the product or if a lot of money that is supposed to be receivable becomes bad debt.

Medical and design (both artistic and mechanical) for sure will be a long term stable AI industry. For consumers I assume assistants. I think there's a good long term market there for AI but I'm not sure about that having anything to do with GPUs. I see AI ASICs taking over eventually for the largest share of the market (enterprise). Having a multiple purpose GPU/CPU that can also do AI is something that will likely only stick around on the consumer side. The power and cost savings of ASICs though are just too much for cloud service providers to ignore. It's a good thing because AI should be extremely accessible.

nguyenNvidia Gaming sector y/y is +18% while AMD is -48% (expected to go down even further in H2 due to low console sale).

I guess Nvidia could just put every research onto their AI super-computers and call it a day

Not good for gamers.

DavenWhat does that mean? Who said anything about discontinuing GPUs?

We are talking about streaming versus local gaming with respect to Nvidia future product and service business strategies. GPUs are still needed for cloud gaming. Lots and lots and lots and lots and lots of them. This conversation comes from the fact that Nvidia data center is 10x the revenue of their gaming category. This could lead them to change, deprioritize or outright exit current ways that they deliver said products and services.

Being able to go to the store and buy a PCIe card called a GPU is totally based on market directions. If you asked someone ten years ago when the optical disc drive of any variety in any device would disappear, most would say never. Same goes for digital non DSLR cameras and even those are close to extinct.

When someone asked me ten years ago whether optical would disappear, and they did, or twenty even, I DID say to them 'DUH, its much better saving the data on something else, like a harddrive, or an SD card, that you can rewrite infinitely. We already had the CD-RW, and it was a hassle. Clear as day that CD was inferior even disregarding the single/multi write issue: burning disks was prone to errors; burnt disks would not be as good as store bought, vibration sensitive players, so even an old walkman with a tape was better for that... prone to damage and wear... etc etc etc. Heck even getting the music on the media was slower when burning to disk, than just copying to a usb stick.

The moment we had mp3, was the moment I was convinced CD or DVD or any other written media was going down the drain. It was easy to predict. Come on. Its just superior in every way to have 'the data' decoupled from the media. And that was long before the Ipod became a thing.

Its not that hard to predict things. Digital photography same thing. Internet, same thing. Sometimes its just obvious that what's new, is going to be simply better. Other times, you'll struggle to draw that conclusion, and that's when you know it ain't going places, or will remain niche at best.

Now, look at cloud gaming and be very honest. How is it worse?

- Its not cheaper. You're simply paying the subscription on top of your games.
- You're not in the same degree of control over your content as you are locally
- You don't control your spending pattern. If Nvidia decides price go up, price go up; whereas locally you can just decide to not buy a game this month. Or buy a 2nd hand GPU. Or stick with your old card.
- If you have no discrete GPU to run bought games, you have a clusterfuck of content you can't access when you aren't paying the service or have it accessible.
- If you don't need to own the games but instead just access the service, you have even less control over your content choices.
- It caters to a specific set/type of gaming, one that doesn't require low latency, meaning it cannot ever replace all gaming, meaning, discrete GPUs will remain in demand.

And how is it better?

- You don't need a discrete GPU to game, so you can play more games in smaller form factors, gaming gains 'mobility'.
- You don't need to upgrade hardware to game, so you can play games in the performance tier the service offers you at price X or Y. So really, instead you're paying your hardware through the subscription. Is this better? Unlikely, but it might just be slightly cheaper over time considering high performance tiers. I'm sure the advantage of scale will reflect a little bit in the price. In fact, Nvidia is already pulling this stunt with NOW. Except... you don't control those priced tiers nor the performance levels on offer.
- ...? Have you got anything else?

Now in the hypothetical event that Nvidia would exit the market, what would happen? They'd leave a multi billion dollar market open to competition. AMD would have no trouble capturing that market now along with Intel. And poof we're back in a duopoly again.