NVIDIA no longer reports gaming as separate revenue line as PCs and consoles move under Edge Computing

[Marees]

NVIDIA puts PCs and game consoles under Edge Computing in latest financial results​

Published: Today, 15:59 GMT • WhyCry

The company now uses two market platforms: Data Center and Edge Computing. NVIDIA says this change “better reflects its current and future growth drivers.” The second category is the one that now catches hardware usually associated with GeForce and gaming systems.

Source: VideoCardz.com
https://videocardz.com/newz/nvidia-...er-edge-computing-in-latest-financial-results

 

[KazeoHin]

And I'll still buy their cards because Nvidia shitting out a half-hearted, overpriced excuse for a product is still better than what AMD would make to compete with it. By AMD's choice.

I would be really surprised, like VERY SHOCKED if AMD didn't follow suit. Every time Nvidia does something terrible, AMD immediately follows them, dick-in hand.

 

[GotNoRice]

I think that there are two big issues at play here-

1. The last financial results where this was categorized separately showed a decline in GeForce / Gaming revenue. This was mainly due to RAM shortages, etc, and of course gaming hasn't been their cash cow for a long time... but Nvidia still seems desperate to hide anything that might be interpreted as "bad news" in any way, shape, or form. Don't want those investors to get scared!
2. Nvidia has so obsessively tried to rebrand themselves as an AI company, it seems like they are beginning to actively disassociate themselves from gaming. Like a political candidate trying to delete all of their old reddit posts, Nvidia would love it now if everyone simply forgot that they were ever associated with gaming.

Maybe it's time for Nvidia to bring back 3DFX and spin that off, so that us plebs can actually have gaming hardware again, and Nvidia doesn't have to be embarrassed about selling it.

 

[littlegreenmenfromhome]

I think that there are two big issues at play here-

1. The last financial results where this was categorized separately showed a decline in GeForce / Gaming revenue. This was mainly due to RAM shortages, etc, and of course gaming hasn't been their cash cow for a long time... but Nvidia still seems desperate to hide anything that might be interpreted as "bad news" in any way, shape, or form. Don't want those investors to get scared!
2. Nvidia has so obsessively tried to rebrand themselves as an AI company, it seems like they are beginning to actively disassociate themselves from gaming. Like a political candidate trying to delete all of their old reddit posts, Nvidia would love it now if everyone simply forgot that they were ever associated with gaming.

Maybe it's time for Nvidia to bring back 3DFX and spin that off, so that us plebs can actually have gaming hardware again, and Nvidia doesn't have to be embarrassed about selling it.

So what is AMD "trying to hide" since they do exactly the same kinda of financial reporting?

Or is this just another FOMO rage with zero thoughts or insigths?

 

[GotNoRice]

So what is AMD "trying to hide" since they do exactly the same kinda of financial reporting?

I don't think that anyone really cares how AMD does or doesn't report their gaming numbers. AMD's gaming GPU marketshare was already low even before the memory scarcity issues hit. While gaming has not been Nvidia's main market for a while, they are still the leader in the field and still have some brand recognition in that area.

Or is this just another FOMO rage with zero thoughts or insigths?

Says the guy who posts the rage post with zero thoughts or insights.

 

[LukeTbk]

it is a lot about not reporting automotive and the others in there own category because of how small they got I suspect (some of those never really cauth up like they thought they would to warrant their own line), having a line for a 600 millions segment in a 80 billions a quarter company start to look strange and probably get them less question about why autonomotive not growing.

Lot of companie will potentially have an edge computing line has well.

 

[El_Capitan]

I would be really surprised, like VERY SHOCKED if AMD didn't follow suit. Every time Nvidia does something terrible, AMD immediately follows them, dick-in hand.

Jensen Huang and Lisa Su are cousins (first cousins once removed). Seems reasonable for AMD to follow NVIDIA's lead, even if the CEO's weren't related, but since they are, it seems more likely that would be the case.

 

[LukeTbk]

So what is AMD "trying to hide" since they do exactly the same kinda of financial reporting?

Usually something similar, when AMD stopped to report gaming operating margin it was after they got really bad a couple of quarters in a row, the first quarter they did it they may have very well been negative, here the non-gaming numbers in that category were getting such a small percentage of Nvidia revenues that they made sense to condense but if automotive was growing well they would have keep it separated, i.e. that not the one that explain the good growth in the edge segment they had.

Here gaming probably grew really well in gaming and pro-viz, but it was becoming "gaming", "pro-vizualisation" quite a bit of many of those GeForce GPU were actually bought to play games versus running other workload could have became hard to tell, same for professional visualization how many of those RTX 6000 series were bought to do solidworks/cad/control room multi monitors room, movie VFX, etc... type of work versus AI.

Could have started to be a bit strange to call those 96 vram gpus VFX industry sales and they got sued in the past for keeping those category in the past when it was hard to tell and not doing what they just did.

Looking at them, that higher than 53.5 billions in GAAP operating income.... that higher than even Apple latest holiday-christmas season quarter operating income and 46% bigger than its regular q2 2026...

Wonder if it is the biggest non Saudi Aramco quarter of all time... when people were saying that Nvidia was massively expensive at 600 billions and too short it, take loan to short it... it was selling at less the double its current ARR and 2.8x it current annualized operating income already... it was one of the cheapest company of all time at those price. (and put in quite the perspective the notion of it being circular revenues when people see less than 2 billions a quarter type of deals if we consider that 100% of nvidia investement in neo-cloud are in pure loop that goes back to hem in comparison to 80+ coming in)

 

[ChadD]

Really don't know why anyone cares how anyone reports their income. What difference does it make exactly?

Jensen told everyone years ago that Nvidia isn't a gaming company. Why do people not listen when he speaks?

They aren't going to stop selling gaming cards. The strategy for 3 generations now has been release the cut down / inexpensive respin of our current thing for the consumer market.
That isn't going to change either. Gamers haven't gotten a full Nvidia chip in a part since the 1080 days. Volta got no consumer release, and since then the biggest fully working silicon has went to datacenter parts.
It really doesn't matter how they report sales numbers to shareholders.

 

[LukeTbk]

Gamers haven't gotten a full Nvidia chip in a part since the 1080 days.

3090ti was the full 10752 core like the RTX A6000, but that was on a samsung node, the A100 type were on TSMC side of things.

I am not sure how popular those RTX 6000 type are popular for datacenter versus going the GB200 option (they probably popular for companie that want something in their own building to use themselve market segment), those were counted in the pro-viz category and now in edge computing that did not explode that much.

The popular datacenter stuff is now all separated from the client you cannot get the cut down version of a GB100 it is a different thing

 

[zehoo]

They hid the crypto and early AI revenue in the gaming revenue. And now they’re hiding their gaming revenue in something else since it must be dropping like a rock.

There must be so much borderline illegal creative accounting going on at nvidia to massage their figures.

The AI crash can’t come soon enough. Will AI go away no, it still has its uses. But the technology to do what they want to do is still 20+ years out. Smoke and mirrors.

 

[El_Capitan]

They hid the crypto and early AI revenue in the gaming revenue. And now they’re hiding their gaming revenue in something else since it must be dropping like a rock.

There must be so much borderline illegal creative accounting going on at nvidia to massage their figures.

The AI crash can’t come soon enough. Will AI go away no, it still has its uses. But the technology to do what they want to do is still 20+ years out. Smoke and mirrors.

Maybe it's to get state tax exemption by keeping AI revenue separate for supplying AI datacenters?

The AI crash will come, most likely in 2028 with a new U.S. President. Much like Obama in 2008, the AI crash will need a bailout like the big banks did, because it's "Too big to fail." This whole AI fiasco is too similar to the 2000's dotcom bubble and the 2008 housing/lending bubble.

How people will react this time around, will be be a different story.

 

[ChadD]

3090ti was the full 10752 core like the RTX A6000, but that was on a samsung node, the A100 type were on TSMC side of things.

I am not sure how popular those RTX 6000 type are popular for datacenter versus going the GB200 option (they probably popular for companie that want something in their own building to use themselve market segment), those were counted in the pro-viz category and now in edge computing that did not explode that much.

The popular datacenter stuff is now all separated from the client you cannot get the cut down version of a GB100 it is a different thing

I'm aware its not the same silicon. I didn't use the best wording when I said the "biggest" fully working. I am saying the full silicon design for the generation goes to the companies real products. Yes they have been making consumer focused versions, which are cut down from the full initial non consumer design, for for some parts are actually fused down larger silicon.
To be fair they are large enough that it makes perfect sense to have 2 or 3 different silicon designs for a generation. Gamers are no longer the market getting access to the best of the gen chips. Which is fine, Nvidia has made clear gaming is a side hustle. A large enough side hustle that they can't actually rely on just using cast off parts to service. [yes I am half joking]

 

[LukeTbk]

I am saying the full silicon design for the generation goes to the companies real products. Yes they have been making consumer focused versions, which are cut down from the full initial non consumer design,

But the full silicon of the RTX 6000 (for which cut down version goes into cheaper RTX5000 and geforce gpus) are not the company real products anymore either (gaming revenues were still much bigger than the pro-viz one)

GB200/datacenter GPU are now the company real product and quite different than a 5070 none of them are fused/cut down/binned down to discrete gpu product, GB200 do not have raytracing cores, do not have ROPs, do not have nvenc/nvdec. GB200 have FP64, encryptions, hardware decompressor, HBM memory controller, NVLink 5 switch.

They do share Tensor cores, Cuda cores and the register design per SM on both, because tensor memory were made for GB200 alone I am not sure how much the cuda-tensor-register size choice needed to be more influenced by the GB200 more than the regular geforce card, the math they do on both are quite similar and the very different memory requirement is handled by something custom present only on the datacenter line.

A bit like having RT workload on RT cores that you simply do not have on datacenter made sure raytracing was not hurting the datacenter business line, that the very large HBM memory controller is only on the datacenter side of things now has well.

At some point it could have looked like that having some much tensor core capability in a SM could be hurting the gaming side of things, but DLSS is such a big success story that not really ended up being the case.

There is a reason a 53,900 millions transistors 9070xt fully enabled GPU on N4P FinFET is not faster than a 45,600 millions on 4N FinFET 5070ti with only 83.333% of the cores enabled, expensive GDDR7 helping bandwidth but also less gaming compromise in the design that some would assume.

 

[LukeTbk]

Maybe it's to get state tax exemption by keeping AI revenue separate for supplying AI datacenters?

datacenter revenues were already their own category, the difference here is putting proviz-automobile-robotics/emmbeded and gaming together into the edge category, a lot of those "provix-gaming" revenues were not really for that anymore, making like 2018 the category more and more misleading.

Before the merge they were:

• Gaming: GeForce GPUs, console chips (like the Nintendo Switch).
• Data Center: A100, H100, Hopper/Blackwell cloud architecture, and Mellanox networking.
• Professional Visualization (Pro-Viz): Quadro and RTX workstation cards (like the RTX 6000).
• Automotive: Self-driving hardware, DRIVE platforms, and infotainment systems.
• OEM & Other: Crypto-mining cards (CMP), legacy parts, and custom low-tier chips.

Now it is data center versus not datacenter revenues together.

This whole AI fiasco is too similar to the 2000's dotcom bubble and the 2008 housing/lending bubble.

If it is similar, it will be interesting to see how big AI will be in 15 years (i.e. people in 2000s underestimated the internet of 2015 by quite a bit), the idea that in 2006/2007 there was too much housing or was being sold a too high price is a bit funny now, housing pricing in 2026 rose from 2006 way faster than inflation, people that bought house at the very top of the market and kept them do not regret it at all now.

The US tax payer made a small profit by buying financial-bank asset at the bottom of the market without even counting the giant good not letting it crash did to them, the bailout of the financial sector did not cost money by the end of it (auto industry that a different story)

The AI crash will come, most likely in 2028 with a new U.S. President. Much like Obama in 2008, the AI crash will need a bailout like the big banks did,

crash and first bailout were under Bush, Obama became the president only in late january 2009, worldwide crisis measure start in 2007 and a 700 billions giant bailout was approved in 2008.

 

[LukeTbk]

Has for the too big to fail around AI datacenter I am not so sure what it would look like.

In 2007-08 the automobile industry employee pension fund and healthcare plan was considered too big to fail and how important cars making was to the global local manufacturing industry and a couple of the big one in GM-Chrysler failing could have made fail part maker that would have made fail the Ford-Toyota-Honda-etc... local plants, loosing all that during the financial-housing crisis would have been a lot.

For the financial-insurance market, a complete credit freeze was of course considered something too big to let happen (same for the insurance world), the amount of corporations that use those very short term loans all the time was vast.

Datacenter currently built world, their failure to be profitable enough to be worth it and a fast cancellation of future project, would it be considered too big to stop ? by who (google, amazon, etc... would still be ok) and well clear the path of a saved GM future (and the banking sector), why save this future if it is discovered to be not worth it ?

I think it would look way more like 1980s investment based on the 70s crisis high petrol price the 2000 dotcom or 1990 telecom crisis than 2008 in term of goverment caring, under (let alone fuitur) datacenter project do not have pension fund/healthcare obligation to millions of well organised unions to lobby the goverment like the auto sector did, not the beloved populace support like the auto industry does. Hyperscaler that are the biggest by far player (microsoft, google, amazon, meta, etc...) are more than solid enough to absorb that shock it is nothing like a GM/Ford debt wise of back in the days. Lot of financing that not from the microsoft direct money is from institutional investor, hedge fund private equity, rich peoples, those goverments tends to not care they knew the risk and can take it, they usually take all the it (they lost everything on GM for example).

Power company that built a lot for future demand if it does not occur via something else would be saved too, same for all the silicon foundry in US soils if a sudden lack of demand put them in trouble, those would be judge (like they are) important enough to be supported regardless of profitability.

Just a question of time for that power/transmission to be useful anyway even without datacenter, USA not making more electricity than 2004 was a terrible state of affair and transmission capacity was making big solar impossible to work well, so you do not want to let it rot and degrade and foundry capacity as obvious national need.

 

[WhoBeDaPlaya]

View: https://www.youtube.com/watch?v=zyQwAhppWj8

 

[littlegreenmenfromhome]

I don't think that anyone really cares how AMD does or doesn't report their gaming numbers. AMD's gaming GPU marketshare was already low even before the memory scarcity issues hit. While gaming has not been Nvidia's main market for a while, they are still the leader in the field and still have some brand recognition in that area.



Says the guy who posts the rage post with zero thoughts or insights.

Deflection, deflection, deflection.

GPU vendors that do not report GPU sales seperataly:
- AMD
-Intel
-NVIDIA

So either A) all are trying to "hide" something or B) you just garbage posts with a bias

Since the burden of "evidence" falls on the one making that claim, ball is in your court.

You attempt of personal attacks proves my point:
You have zero data and bias post.

 

[littlegreenmenfromhome]

Usually something similar, when AMD stopped to report gaming operating margin it was after they got really bad a couple of quarters in a row, the first quarter they did it they may have very well been negative, here the non-gaming numbers in that category were getting such a small percentage of Nvidia revenues that they made sense to condense but if automotive was growing well they would have keep it separated, i.e. that not the one that explain the good growth in the edge segment they had.

Here gaming probably grew really well in gaming and pro-viz, but it was becoming "gaming", "pro-vizualisation" quite a bit of many of those GeForce GPU were actually bought to play games versus running other workload could have became hard to tell, same for professional visualization how many of those RTX 6000 series were bought to do solidworks/cad/control room multi monitors room, movie VFX, etc... type of work versus AI.

Could have started to be a bit strange to call those 96 vram gpus VFX industry sales and they got sued in the past for keeping those category in the past when it was hard to tell and not doing what they just did.

Looking at them, that higher than 53.5 billions in GAAP operating income.... that higher than even Apple latest holiday-christmas season quarter operating income and 46% bigger than its regular q2 2026...

Wonder if it is the biggest non Saudi Aramco quarter of all time... when people were saying that Nvidia was massively expensive at 600 billions and too short it, take loan to short it... it was selling at less the double its current ARR and 2.8x it current annualized operating income already... it was one of the cheapest company of all time at those price. (and put in quite the perspective the notion of it being circular revenues when people see less than 2 billions a quarter type of deals if we consider that 100% of nvidia investement in neo-cloud are in pure loop that goes back to hem in comparison to 80+ coming in)

GPU vendors not report GPU sales seperatly:

- AMD
- Intel
- NVIDIA.

A big nothing burger.

 

[LukeTbk]

A big nothing burger.

Change can often be read as a sign (take the recent AMD stopping operating margin of the gaming division)

Here it is a lot so relatively small for auto/others that it made less and less sense to have their own line but also as being sued in the past for this, the name of the category was getting fuzzier and fuzzier with actual use (and we can read into it that automotive was never picking up, which is a bit of nothing burger true).

One can reasonably assume that their telemetry and other sign/feedback offf reseller was telling them that the percentage of those GPU being used for gaming and pro-viz work were getting dangerously low, what percentage of RTX 6000 were really going into workstation and not for AI type work...

It is a bit opposite to try to hide something (and what people seem to point out, that they were hiding previously AI revenues into gaming and pro-viz)

 

[littlegreenmenfromhome]

Change can often be read as a sign (take the recent AMD stopping operating margin of the gaming division)

Here it is a lot so relatively small for auto/others that it made less and less sense to have their own line but also as being sued in the past for this, the name of the category was getting fuzzier and fuzzier with actual use (and we can read into it that automotive was never picking up, which is a bit of nothing burger true).

One can reasonably assume that their telemetry and other sign/feedback offf reseller was telling them that the percentage of those GPU being used for gaming and pro-viz work were getting dangerously low, what percentage of RTX 6000 were really going into workstation and not for AI type work...

It is a bit opposite to try to hide something (and what people seem to point out, that they were hiding previously AI revenues into gaming and pro-viz)

Perhaps they are trying to hide that the 5080's share on Steam is more than double the 4080? (Steam Data)
Or that the 5090 is tracking faster than the 4090 did? (Steam Data)
Or that they have gotten 94% of sales in the AIB private market? (Jon Peddie data)

For one to make the evidence‑averse claims that are being made here, one has to ignore the available data, because it conflicts with bias and FOMO.

Just like the A.I. haters posting from feeling and not facts (NIVIDA 85% increase in revenue YoY Q1)

A big nothing burger fueled by stupid.

 

[LukeTbk]

Perhaps they are trying to hide that the 5080's share on Steam is more than double the 4080? (Steam Data)

4080 super split cards sales mid cycle, 1.44% is excellent for a 5080 of course and even a bit higher than the 4080/4080 super of the survey (1.38%)

I am not sure what the claim is in term of hiding, would they hide something by the way they merged things, would be automotive sector never picking up and just continue to hide how much of the pro-viz/gaming business is an AI one now (not they can know for sure, but with high percentage of gamers using nvidia apps, they get good telemetry data on that one at least), which they were of course already doing before this announcement for that one, so this change is not for hiding it, if anything they are more transparent about this.

according to the steam survey already ~23% of world gamers are on discrete blackwell gpu, 28% with blackwell Laptops.

Thats could be the best-close to the best selling gen just for gaming use ever after just what 15-16 months, with 180 millions accounts that ~41.4 millions GPU, 5 quarters with around 9-12 millions dgpu a quarter goes out I think, so that seem in pace.

 

[GotNoRice]

Deflection, deflection, deflection.

I understand that you are confused. I'll spell it out with a crayon for you.

GPU vendors that do not report GPU sales seperataly:
- AMD
-Intel
-NVIDIA

AIB GPU Marketshare:

AMD: 5%
Intel: 1%
NVIDIA: 94%

Most of us (including, you know, the actual article) are discussing the Monopoly with 94% marketshare that dominates and controls the market. You're waving your finger at the 1% and 5% companies and claiming "they do it too!!!!!", and somehow don't understand why it doesn't matter.

You attempt of personal attacks proves my point:

My only "attempt of personal attack" was literally repeating back to you exactly what you posted. If you really find your own content so offensive, take solace in knowing that at least you've identified the problem.

 

[littlegreenmenfromhome]

I understand that you are confused. I'll spell it out with a crayon for you.



AIB GPU Marketshare:

AMD: 5%
Intel: 1%
NVIDIA: 94%

Most of us (including, you know, the actual article) are discussing the Monopoly with 94% marketshare that dominates and controls the market. You're waving your finger at the 1% and 5% companies and claiming "they do it too!!!!!", and somehow don't understand why it doesn't matter.



My only "attempt of personal attack" was literally repeating back to you exactly what you posted. If you really find your own content so offensive, take solace in knowing that at least you've identified the problem.

Can you find your crayons and draw for me some facts as I have zero interest in your conjecture, hurt feeling or personal attacks.

 

[littlegreenmenfromhome]

4080 super split cards sales mid cycle, 1.44% is excellent for a 5080 of course and even a bit higher than the 4080/4080 super shader of the survey (1.38%)

I am not sure what the claim is in term of hiding, would they hide something by the way they merged things, would be automotive sector never picking up and just continue to hide how much of the pro-viz/gaming business is an AI one now (not they can know for sure, but with high percentage of gamers using nvidia apps, they get good telemetry data on that one at least), which they were of course already doing before this announcement for that one, so this change is not for hiding it, if anything they are more transparent about this.

according to the steam survey already ~23% of world gamers are on discrete blackwell gpu, 28% with blackwell Laptops.

Thats could be the best-close to the best selling gen just for gaming use ever after just what 15-16 months, with 180 millions accounts that ~41.4 millions GPU, 5 quarters with around 9-12 millions dgpu a quarter goes out I think, so that seem in pace.

I do not see them hiding anything their marketshare is growing into total dominance on both the consumer and enterprise side.
The only thing "abnormal" is the vocal minorty crying "foul" (with nothing but bias and conjecture) as the vast majority is saying "shutup and take my money" (due to their products haveing the perferred performance/features) hence the debate is twisted in a way that does not match reality.

 

[gtrguy]

Talk about hurt feelings LOLOLOLOL

 

[littlegreenmenfromhome]

Talk about hurt feelings LOLOLOLOL

Yes, the 5% seems very hurt

 

[Lakados]

I’m surprised it took Nvidia this long.

When they were getting sued over Gaming GPU sales numbers propped up by Crypto mining, it became the obvious answer. Now with AI chewing up GPU’s even faster than Crypto did it’s a necessity.

Nvidia doesn’t track what you do with the cards, just because they internally designate the GPU for Gaming, or Mining, or AI doesn’t mean that’s what it will be used for.

So if Nvidia states that all 5080 and 5090 GPU sales were gaming sales, then some time later it’s discovered that some significant portion of those were actually AI sales someone is going to lawyer up over it hoping for a fast settlement.
It happened with Crypto, it will happen with AI.

This move just clean up a technicality that Nvidia could find itself sued over, especially if the AI bubble bursts.

 

[zehoo]

Nvidia doesn’t track what you do with the cards, just because they internally designate the GPU for Gaming, or Mining, or AI doesn’t mean that’s what it will be used for.

Unless you turn telemetry off in the drivers (which isn’t an easy toggle but editing the registry), I’m pretty sure they track what you do with the card. In windows anyway.

 

[Lakados]

Unless you turn telemetry off in the drivers (which isn’t an easy toggle but editing the registry), I’m pretty sure they track what you do with the card. In windows anyway.

That data isn't detailed enough to be useful in a lawsuit against Nvidia, and they keep it that way.
In Windows, if you do just the Driver install, then the logging functionality is limited to driver issues and GPU usage metrics; it's pretty minimal. You don't have to install the Nvidia Experience BS, and without it, most of the reporting tools don't ever get installed (it's why Nvidia pushes it so hard)
Besides, most of the people doing AI, Crypto, etc., aren't doing it in Windows, so the usage logs skew heavily towards them being used primarily for gaming, which doesn't help the lawyers argue that Nvidia "knew" that the sales weren't going to gaming machines.