AMD Radeon RX 9000 GPUs begin to appear in the Steam Hardware Survey at last — RX 9070 arrives with paltry 0.16% market share, less than the GeForce GT 730
There has also been a significant surge in folks packing 16GB VRAM GPUs.
The latest Steam Hardware Survey is out, and an AMD RDNA4 architecture GPU has charted in the PC Video Card Usage tables. Despite AMD’s fanfared deprioritization of flagship GPUs to follow the mass market, the Radeon RX 9000 family's entry into the chart has been a long time coming. Moreover, the RDNA4 Radeon has appeared with more of a whimper than a bang, turning up with just a 0.16% market share, still less than Nvidia's 2014 GeForce GT 730.
Your eyes don’t deceive – AMD’s only Steam Hardware Survey (SHWS) charted card is the RX 9070 non-XT. Does this show that AMD’s cheaper, slightly lower-tier model may be a sleeper hit for its amazing efficiency? I have a plain vanilla RX 9070 in my personal desktop, and am very pleased with its quiet, cool performance, so praise of this SKU might show a little bias.
Interestingly, though, AMD’s RX 9070 XT did indeed hit the SHWS in the December 2025 data, at 0.22%. It has now disappeared from view (under the 0.15% threshold for charting).
It has been approaching a year since the first RX 9000 graphics cards became available. We reviewed the RX 9070 XT and RX 9070 back in early March 2025. These 16GB VRAM cards were praised for strong mainstream performance at good MSRPs, with worthwhile improvements in AI and ray tracing performance. AMD released its RX 9060 XT (8 and 16GB VRAM) models in June last year, and also saw largely positive feedback (for the 16GB version, anyway).
It is worth repeating that, despite AMD’s mass market strategy and general positivity in reviews and on discussion forums, people are still buying into the GeForce ecosystem by default.
While we are kind of celebrating the first RDNA4 card charting in the Steam usage database with its 0.16% placing, the last-gen RTX 4060 gained 0.46% share. Moreover, the biggest gainer of the month was, predictably, the RTX 5060 with an increase of 0.72% in market share. If we sort the chart by gains in the last month, the RX 9070 rises from near the bottom of the chart to 6th place. Which isn’t so bad.
16GB VRAM adoption spurt
Another change that seems remarkable in the last month is the observed increase in Steam gamers backing GPUs with 16GB of VRAM. As per the numbers shown, those gaming on 16GB GPUs went up an impressive 5.85% in January 2026. Meanwhile, 12GB VRAM became 4.01% less prevalent, and 8GB VRAM, 3.11% less common.
Get Tom's Hardware's best news and in-depth reviews, straight to your inbox.
So, 16GB of VRAM looks set to move into second place in the graphics card memory charts, as early as February’s results. Unless there is another inflection or bump in the trends...
Last but not least, please remember that the SHWS figures aren’t definitive, as the disappearance of the RX 9070 XT shows. They are just an indicator of trends among active Steam gamers, snapshotted that month. Sometimes big events in the Steam marketplace, like a country being added to the survey, can nudge the results more than usual, and the way the charts are compiled is quite opaque. Other factors, like seasonality and big game releases, can also impact the data.
Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.
Mark Tyson is a news editor at Tom's Hardware. He enjoys covering the full breadth of PC tech; from business and semiconductor design to products approaching the edge of reason.
-
Jabberwocky79 How does the 9070XT disappear from the list? Does Steam throw out results from previous surveys and only show a current snapshot of the random users that were surveyed in the current round? I always thought of the hardware survey as an aggregate, but maybe that isn't the case... that's the only way for these numbers to make sense.Reply -
edzieba Reply
It's a monthly sampling. Valve are entirely opaque as to how that sampling actually works (moving average window size, percentage of Steam population that gets surveyed, weighting of sampling by region, etc).Jabberwocky79 said:How does the 9070XT disappear from the list? Does Steam throw out results from previous surveys and only show a current snapshot of the random users that were surveyed in the current round? I always thought of the hardware survey as an aggregate, but maybe that isn't the case... that's the only way for these numbers to make sense. -
LordVile Reply
You are talking about a company that can’t count to 3Jabberwocky79 said:How does the 9070XT disappear from the list? Does Steam throw out results from previous surveys and only show a current snapshot of the random users that were surveyed in the current round? I always thought of the hardware survey as an aggregate, but maybe that isn't the case... that's the only way for these numbers to make sense. -
DS426 SHWS data is certainly limited in value. 1 GB, 2 GB, 3 GB, and 4 GB VRAM GPU's all had gains in January. Nice.Reply -
Jabberwocky79 Reply
:ROFLMAO:DS426 said:SHWS data is certainly limited in value. 1 GB, 2 GB, 3 GB, and 4 GB VRAM GPU's all had gains in January. Nice. -
salgado18 Reply
I really respect Valve for all that it is and has achieved, but I also really doubt they use actual statistics concepts for these surveys. It feels more like "ok, lets get some random users and do some math with them, because reading everyone, saving their hardware, and monitoring for changes is too much work".edzieba said:It's a monthly sampling. Valve are entirely opaque as to how that sampling actually works (moving average window size, percentage of Steam population that gets surveyed, weighting of sampling by region, etc). -
Shiznizzle Reply
I dont understand why the survey has to be voluntary. Make everybody submit. At least then the numbers are reflective of what is happening on the platform as opposed to some random number of users who submitted based upon some RNG chance of getting the invite.salgado18 said:I really respect Valve for all that it is and has achieved, but I also really doubt they use actual statistics concepts for these surveys. It feels more like "ok, lets get some random users and do some math with them, because reading everyone, saving their hardware, and monitoring for changes is too much work".
unrelated. "Moreover, the RDNA4 Radeon has appeared with more of a whimper than a bang, turning up with just a 0.16% market share, still less than Nvidia's 2014 GeForce GT 730."
yeah but my 9060XT 16GB puts that GT 730 to shame. Was wondering why my card was not showing up 4 or more months now after i bought it -
thestryker Reply
The hardware survey is a current extrapolation. The biggest issue here is that Valve doesn't have any sort of restrictions on the data pool. Hypothetically I could see one survey a year and you could see one each month. This is how we end up seeing spikes of varying sorts randomly appearing over the years.Jabberwocky79 said:How does the 9070XT disappear from the list? Does Steam throw out results from previous surveys and only show a current snapshot of the random users that were surveyed in the current round? I always thought of the hardware survey as an aggregate, but maybe that isn't the case... that's the only way for these numbers to make sense.
They use proper statistics methodologies for the survey it's just the data itself is suspect. As long as the hardware survey is a random, voluntary, sampling without ensuring the input data is good it's of limited use. -
mormodra I have a 9070XT and I am not that special... so I imagine those numbers are significantly higher.Reply -
Treggs Reply
Yeah its is significantly higher what the media always fail to mention or probably dont even do the research to learn is, if you have an x3d chip with intergrated gpu and you have a 9000 series gpu then because of a bug in the steam servey the igpu shows up instead of the 9000 series card.mormodra said:I have a 9070XT and I am not that special... so I imagine those numbers are significantly higher.