Tuesday, December 2nd 2025

AMD Radeon RX 9000 Series Still Absent from Steam Hardware Survey

As we enter December, Valve has completed processing data and officially released the November Steam Hardware and Software Survey, highlighting trends in the gaming industry. Interestingly, AMD's latest RDNA 4 designs from the Radeon RX 9000 series have yet to make a significant impact on the market, indicating that many Steam survey participants are not using AMD Radeon RX 9000 series GPUs. The top spot in the GPU usage list is held by NVIDIA's GeForce RTX 4060 (notebook), with the first AMD product appearing nine spots later. Valve lists this first AMD Radeon GPU as an unnamed entry, likely representing integrated GPUs or other systems-on-chip products with unconventional configurations compared to standard discrete GPUs.

The first AMD discrete GPU on the list is the older RDNA 2-based Radeon RX 6600, followed by the RDNA 3-based RX 7800 XT. However, nine months after the RDNA 4 series began selling, there are no RDNA 4 GPUs present in the general survey. This could suggest that Steam gamers prefer other options or that the RDNA 4 series isn't selling well. While we don't have a definitive conclusion, the data points to a concerning trend.
When ranking GPU popularity on Linux operating systems, the AMD Radeon RX 9070 XT comes in at 17th place. This isn't ideal market penetration for a lineup that has been available for nearly a year. The existing and upcoming Linux based hardware from Valve itself still only use RDNA 2 and RDNA 3 based graphics. However, since most systems run on Windows, the presence of RDNA 4 GPUs on Linux-based OS is still negligible in the overall market. Hence, it seems that the RDNA 4 generation will take at least a few more months before Steam Survey shows any dent in the market data.
Source: Steam Survey
Add your own comment

75 Comments on AMD Radeon RX 9000 Series Still Absent from Steam Hardware Survey

#51
LastDudeALive
Red HoodYou mean where they said the steam machine is equal or better than 70% of pcs in the survey? To add to my previous comment, they could collect the pc specs from all the steam clients (that are connected to the internet) which would be a lot more accurate than results from random surveys?
No, actually. The current method requires an actual person sitting in front of the computer to say "yes." If they just polled all steam clients, it would give non-active users and other irrelevant data more weight in the survey. There's also the significant privacy concerns, especially for places like the EU. A lot of data is gathered from the survey. Taking it all without consent, or burying the consent in the standard TOS of the Steam client itself, would probably attract unwanted legal attention.

Steam only needs to survey ~20,000 random people every month to have a statistically sound data set. They probably prompt significantly more than that for the survey, to make sure they get enough data for a sound result.
Posted on Reply
#52
wolf
Better Than Native
HereticbarLet me get his straight. So Valve is a member of a secretive cabal that is actively engaging in rigging their own survey to undercount AMD GPU usage numbers among Steam users. And this is done with malicious intent to create the impression that AMD GPUs are inferior and less desirable.

While at the same time, all current and upcoming Steam hardware they are selling uses Radeon GPUs.

Got it.
Don't hear many complaints about the rise and share of AMD CPU's on the survey, Radeon seems to be the focal point of the persecution complex, the only difference here is not hearing that it's somehow nGReEeDIA's fault.
Posted on Reply
#53
Hecate91
HereticbarLet me get his straight. So Valve is a member of a secretive cabal that is actively engaging in rigging their own survey to undercount AMD GPU usage numbers among Steam users. And this is done with malicious intent to create the impression that AMD GPUs are inferior and less desirable.

While at the same time, all current and upcoming Steam hardware they are selling uses Radeon GPUs.

Got it.
I don't see where anyone had said anything of the sort in this thread, just GeForce users in this thread looking to mock AMD users as usual while jumping to conclusions to defend the beloved survey because it puts the favorited brand on a pedestal.

Then again this is free or very cheap marketing for Nvidia because all it takes is news sources posting survey results, interestingly enough the 5070 gets praised for replacing the 4070 in the survey, I guess enough people fell for the "4090 performance" false advertisement.
Posted on Reply
#54
wolf
Better Than Native
Hecate91just GeForce users in this thread looking to mock AMD users as usual
You see what you want to see.
Hecate91while jumping to conclusions to defend the beloved survey because it puts the favorited brand on a pedestal.
Again seeing it how you want to see it, it's about statistics, not about putting a brand on a pedestal. It's easier to just paint people that way though, then you can write off anything they say as irrelevant.
Hecate91interestingly enough the 5070 gets praised for replacing the 4070 in the survey, I guess enough people fell for the "4090 performance" false advertisement.
Interesting leap of logic there to conclude that the 5070 selling well is because significant numbers of people saw that one slide, completely believed it and bought the 5070 just for that reason.
Posted on Reply
#55
londiste
OnasiIs the survey broken or does AMDs driver present incorrect information to the system? The survey isn’t doing anything special by itself - it merely polls the OS. It cannot be broken by its very nature. It picks up what is presented. The issue is either on AMDs end or MSs.
The driver description string for AMD cards showed the generic "AMD Radeon Graphics" for a long while for a lot of different and proper cards. I really thought AMD changed this quite a few generations of GPUs back now. From this specific topic the screenshots in this thread and elsewhere seem to point at survey properly detecting 9070s. In recent past there has been conflicting information though about whether survey is able to discern 9070XTs from 9070s.

An argument could be made that survey could or should do some more advanced detection other than taking the driver string but counter-argument that this should not be necessary is as valid.
Posted on Reply
#56
Visible Noise
ThomasKGuess Valve is happy not asking actual RDNA4 owners to submit their specs for the hardware survey. I've had my 9070 XT practically since launch, and haven't got a single submission request from them.
I have got to know how in the fuck you think Valve isn’t asking RDNA4 owners to take the survey, when they don’t know what hardware the client is running on unless you take the survey. Valve doesn’t even know who the owner of a computer is. Computers get scanned, not accounts.

What’s in it for Valve?
Hecate91Yet once in a while Valve still allows a massive influx of gaming cafes to influence the whole survey, it simply isn't trustworthy
You mean right after the Chinese new year?

Gee I wonder why a lot of businesses that have been closed for weeks suddenly show a jump in computer usage. Can you think of any reason why it happens at the same time every year?
Posted on Reply
#57
Shihab
Red HoodSo we don't know how accurate the steam survey is, hence the doubts in the validity of it's results
You don’t need to know the exact sample size to figure out that it’s large enough or not. The mere fact that the timeseries doesn’t look like the waveform of a Skrillex track is indicative of that.
Red Hoodwhich would be a lot more accurate than results from random surveys?
The relationship between accuracy and sample size is logarithmic. If your population is a million, jumping from ten samples to a hundred would give you a huge boost in accuracy, but going from 10k to the whole million wouldn't even add 0.1% accuracy (probably).
TheinsanegamerNIf you survey all 20,000 users in China, or in Russia, you won't have an accurate representation.
If, for the sake of argument, we agreed on the belief that the total of Chinese and Russian users don't make the bulk of Steam's user base, the probability of 20k out of 20K being all Chinese or Russian is so low, I don't think the digital data types we have can even encode it!:roll:

For the record, I actually did try to compute it (a naive hypergeometric distribution calculator), and I misread LastDudeAlive's 70 mil for 7 mil (with hypothetical, 50% user base as Chinese). The script still froze for 10 minutes before spitting out a big zero...
Posted on Reply
#58
Scircura
TheinsanegamerNPolling every user once a month is trivial, and would give 100% accurate data. Valve doesnt have to send people out i the street sto collect this,e very steam client already has this data, it just needs to be uploaded. I'm no mathematician, but 100% is more accurate then any number under 100% last time I checked.
The survey is opt-in. Valve does not, and should not, automatically collect detailed information on everyone's machines, because that's creepy and probably violates many data protection laws. So Valve must ask you to participate in the survey.

Nagging every Steam user every month with a survey will irritate and be ignored by all but the sweatiest tech enthusiasts who desperately want Valve to know how sweet their rig is. So the survey results will over-represent 5090s and 240Hz OLEDs, and may be less accurate than an extrapolated result from a 1% sample.

(This example will be included in my upcoming book, with the working title 101 Surprising Facts About Surveys.)

And I won't even try to answer the "debate" about survey sampling in this thread. But in general, folks should start from the position that Valve knows what it's doing, and think carefully before challenging that.
Posted on Reply
#59
Hereticbar
Hecate91Then again this is free or very cheap marketing for Nvidia because all it takes is news sources posting survey results, interestingly enough the 5070 gets praised for replacing the 4070 in the survey, I guess enough people fell for the "4090 performance" false advertisement.
So every news site that accurately report the results of a hardware survey done by Steam is now being seen as providing "free or very cheap marketing for Nvidia"?

How many layers deep does is this grand conspiracy goes? Steam is in on it obviously, and news sites are all in the take too. :kookoo:

Jensen Huang and Nvidia don't care about gamers or the gaming market in general, all they care about is AI. However at the time they care so such about the gaming market, they spared no effort to corral Value into their evil scheme of rigging the Steam Hardware survey in their favor and ruthlessly suppress the 1% of users that might show up as using RX 9000 series GPUs.

Which one is it? Make up your damn minds.
Posted on Reply
#60
Unregistered
LastDudeALiveYou want to poll randomly
Going by that parameter, imagine the perfect storm that only Ngreedia users are pooled, is that survey still correct?

And yes, given how you and the others love to debate, technically, the survey is correct because its showing the collected data, but no, it is flawed.
LastDudeALiveThey only need to poll ~20,000 random people every month
Again, going by the perfect storm situation, if all of those are Ngreedia GPU's, its that dataset valid?
W1zzardGiven their market share, this must be the case to be fair?
See the points above.


As mentioned, the Steam Client reports your hardware back to Valve because it needs to know if your system is compatible with it.

From Valve:
  • Data collected: The client uses your hardware information to determine if you meet the minimum system requirements for a particular game.

    But I know, the debate must go on, the loyal Ngreedia white (green?) knights must defend the honor of that poor and defenseless corporation and we all should be grateful of enjoying a monopoly.
#61
mechtech
Availability and availability at MSRP has been a issue.............

With our dollar now 1.40 to 1 usd the exchange rate doesn't help and with the price of video ram going up that won't help either.

Looks like the old 6800 will be going for at least another 3 years.
Posted on Reply
#62
Red Hood
store.steampowered.com/privacy_agreement/

3.5 Your Use of Games and other Subscriptions
In order to provide you with services, we need to collect, store and use various information about your activity in our Content and Services. "Content-Related Information" includes your Steam ID, as well as what is usually referred to as "game statistics". By game statistics we mean information about your games' preferences, progress in the games, playtime, as well as information about the device you are using, including what operating system you are using, device settings, unique device identifiers, and crash data.
Posted on Reply
#63
Shihab
Neo_MorpheusGoing by that parameter, imagine the perfect storm that only Ngreedia users are pooled, is that survey still correct?

And yes, given how you and the others love to debate, technically, the survey is correct because its showing the collected data, but no, it is flawed.
That's not what it means and your hypothetical makes no sense.
There also exists a "perfect storm" where a 100, heavy meteorites fall on your head and 99 people around you.
Neo_MorpheusAs mentioned, the Steam Client reports your hardware back to Valve because it needs to know if your system is compatible with it.
This is a strawman, but sure. Compatibility check is done locally as a standard. That’s why every hardware-dependant API exposes tonnes of calls to check supported features (Ask OpenGL/Vulcan devs how fun it is to check for extensions).

Now, hypothetically, if Valve does have the exact same data types as the survey but for all users and they still elect to use and publish the opt-in ones, you have two positions you have to pick one of:
A- Claim Valve is intentionally publishing false data.
B- Accept that Valve deems the opt-in data no different than the this hypothetical all-encompassing one.

B would be easier on everyone’s sanity, but I will not deny that I’m interested to see the mental gymnastics needed to rationalise A.
Posted on Reply
#64
GodisanAtheist
I dunno why it's so hard to understand that 9070xt is selling well does not equal 9070XT is outselling Nvidia, or is selling enough to show up on the survey.

AMD is completely AWOL in laptops, barely has a presence in OEMS, and is still outsold multiple to one in the DIY market.

Just because you bought a 9070xt doesn't mean 10 other people didn't just go out and buy an Nvidia card instead, even as part of a pre built or laptop.

And I say this as a 6800xt owner with no beef toward AMD whatsoever.
Posted on Reply
#65
LastDudeALive
GodisanAtheistI dunno why it's so hard to understand that 9070xt is selling well does not equal 9070XT is outselling Nvidia, or is selling enough to show up on the survey.
Exactly. The only concrete "numbers" AMD have provided is that RDNA 4 outsold previous launches by a factor of 10, which I fully believe, when you compare them to Nvidia in previous generations (6800 XT vs 3080/3090, 4090/4080 vs 7900 XTX/XT). But that doesn't mean they're anywhere close to being in Nvidia's ballpark, it just means their previous generations really sucked.
Posted on Reply
#66
Unregistered
ShihabThat's not what it means and your hypothetical makes no sense.
There also exists a "perfect storm" where a 100, heavy meteorites fall on your head and 99 people around you.


This is a strawman, but sure. Compatibility check is done locally as a standard. That’s why every hardware-dependant API exposes tonnes of calls to check supported features (Ask OpenGL/Vulcan devs how fun it is to check for extensions).

Now, hypothetically, if Valve does have the exact same data types as the survey but for all users and they still elect to use and publish the opt-in ones, you have two positions you have to pick one of:
A- Claim Valve is intentionally publishing false data.
B- Accept that Valve deems the opt-in data no different than the this hypothetical all-encompassing one.

B would be easier on everyone’s sanity, but I will not deny that I’m interested to see the mental gymnastics needed to rationalise A.
Goalpost meet goalpost mover....

Honestly, its useless having any kind of conversation with certain groups of people.
#67
Tomorrow


18,05% by Steam itself. Not 0%, Not 7%.
And deducting iGPU and mobile cards actually increases that to over 19% because AMD is pretty much nonexistent in Steam charts when it comes to mobile GPU's. However deducting those from Nvidia share actually makes AMD number go up.

And this is from Steam that is already suspected of under reporting AMD GPU's.
Current threshold of showing up in the main table (not just DX10/11/12 & Vulkan) is 0,16%.
9070 XT is currently at 0,13% so it will likely show up in the main table sometime in the first half of next year.
Posted on Reply
#68
Shihab
Neo_MorpheusGoalpost meet goalpost mover....

Honestly, its useless having any kind of conversation with certain groups of people.
Which group of people? Those little, green men that got you so paranoid? Sorry for the probes. :D
Posted on Reply
#69
LastDudeALive
TomorrowAnd deducting iGPU and mobile cards actually increases that to over 19% because AMD is pretty much nonexistent in Steam charts when it comes to mobile GPU's. However deducting those from Nvidia share actually makes AMD number go up.
Why should we exclude mobile GPUs just because AMD can't be bothered to compete in that market? Their absence should hurt AMD's marketshare. Mobile GPUs are just as important as discrete GPUs.
Posted on Reply
#70
PixelTech
I had my 3rd or 4th survey submission happen on Monday, 12/1.
Posted on Reply
#71
wolf
Better Than Native
Neo_MorpheusHonestly, its useless having any kind of conversation with certain groups of people
You're describing yourself, but with extra steps.

Sometimes it feels like I'm in a home theatre forum with the amount of projection going on.
Posted on Reply
#72
Ssenol
AMD's indecisiveness regarding driver development has shaken user trust. Its announcement of maintenance mode for its RDNA 2 cards, which are less than five years old, has led to a flood of people selling AMD cards, demanding they never buy AMD again! Despite having an INT8 version of FSR 4, it stubbornly refused to release it to previous-generation cards, and withheld feature updates very early on just to avoid releasing FSR 4 to older-generation cards. Who can trust AMD anymore? Will they buy new cards?
Posted on Reply
#73
cvearl
Interresting how, in the last quarter, there has been almost a 4% increase in PC's with installed AMD GPU's in the Survey! Going from under 15% to over 18% in the last 4 or 5 months and no 9070XT to speak of. Yet there are now 3 separate generic AMD GPU's listed as 3 separate line items. That used to be only 2 for yaers. I honestly think STEAM is suppressing AMD 9070XT reporting by accident or on purpose regardless of what anyone says.
SsenolAMD's indecisiveness regarding driver development has shaken user trust. Its announcement of maintenance mode for its RDNA 2 cards, which are less than five years old, has led to a flood of people selling AMD cards, demanding they never buy AMD again! Despite having an INT8 version of FSR 4, it stubbornly refused to release it to previous-generation cards, and withheld feature updates very early on just to avoid releasing FSR 4 to older-generation cards. Who can trust AMD anymore? Will they buy new cards?
This was walked backed soon after so... ya. They are still supporting them.
Posted on Reply
#74
LastDudeALive
cvearlYet there are now 3 separate generic AMD GPU's listed as 3 separate line items.
They're iGPUs, AMD should standardize how their driver reports the iGPUs, or break it out further (for instance, it seems on some systems 780M GPUs are individual reported).
Posted on Reply
#75
Ssenol
cvearlInterresting how, in the last quarter, there has been almost a 4% increase in PC's with installed AMD GPU's in the Survey! Going from under 15% to over 18% in the last 4 or 5 months and no 9070XT to speak of. Yet there are now 3 separate generic AMD GPU's listed as 3 separate line items. That used to be only 2 for yaers. I honestly think STEAM is suppressing AMD 9070XT reporting by accident or on purpose regardless of what anyone says.


This was walked backed soon after so... ya. They are still supporting them.
They indicated they were withdrawing it, but there's still nothing new, no corrections or additions. In short, it remains just a promise.
Posted on Reply
Add your own comment
Nov 21st, 2026 13:52 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts