Steam survey sees a shake-up as a new top graphics card is revealed

midian182

Posts: 11,722   +177
Staff member
What just happened? There's a new top graphics card on the Steam survey. After seeing a comparatively huge 4% increase in users, the RTX 4060 has replaced the RTX 3060 as the most popular GPU among participants of Valve's survey. There was also an unexpected 5% jump in the number of people using Intel CPUs, which points to February being another one of those months where the results were anomalous.

February saw the user share for both the RTX 4060 and RTX 4060 Ti jump by 3.97% and 3.11%, respectively. That's a huge increase compared to the less than 1% changes we usually see each month.

The rest of the top 12 performers of the month are made up of variants of RTX xx60 and RTX xx70 cards from the Lovelace, Ampere, and Turing generations.

The RTX 4060 has been catching up to the RTX 3060 for a while now, so it's not too surprising that it's taken the lead, but leapfrogging its predecessor with a 4% increase is unusual. As with the top performers, the most popular cards are made of xx60 and xx70 GPUs.

February also saw the overall number of Nvidia GPUs on the table increase. Team Green now accounts for 83% of products on the list. AMD has 11.5%, and Intel has 5.2%.

Another strange result was in the CPU section. AMD has spent months eroding Intel's lead, with Team Red hitting a record 36.19% share in January. But February saw AMD fall 5% as Intel rose by the same amount – a contrast to what we've seen in the retail space this year.

In further evidence that this is one of those weird Steam survey months, Windows 10, which had fallen below Windows 11 as the most-used OS among participants, suddenly retook the top spot after its share skyrocketed by 10.5% as Windows 11 dropped just over 9%.

Windows 10 reaches its end of support date on October 14, 2025. According to Statcounter, its global user share has dropped over the last two months, from 62.7% in December to 58.7% in February, while Windows 11 has seen its share climb.

Elsewhere on Valve's survey, 32GB suddenly became the most popular amount of system RAM following a 13.7% gain and those using 16GB fell by 8%. There's also a new most-popular language, Simplified Chinese, which saw its usage go up 20% to take a 50% overall share as English fell 10%.

There was similar strangeness in the survey results back in October 2023, with unusually large changes in a lot of categories, including Chinese going up almost 14%. Things returned to normal a month later, so March's survey could look very different.

Permalink to story:

 
What?? Surely you'd notice its an outlier where after months of no changes, cards suddenly get multi percent uplifts for cards of the *old* generation, the only newsworthy thing about this is that there was probably a big intake of data from somewhere and it should be accounted for
 
All you need is a single stat to see this month's Steam Survey is even more complete WHACK than it usually is:

Simplified Chinese 50.06%+20.88%
English 23.79% -10.18%
Russian 6.76% -2.86%

Yes, the Chinese userbase expanded by 70% while English and Russian contracted by 30%. What a f***ing joke.
 
The survey is only a moment in time, and where the steam bots poll a system (If you allow). So the bots were busy in China this month. (plus I could see where the polling numbers could easily be hacked.) The poll is not gospel for anything. However, over months it may show general trends. If you follow the poll regularly, you should expect hardware and OS numbers to bounce back as steam polls other regions next month - they always do.

This months bounce for Win 10 is funny, I bet we'll see an opposite bounce for Win 11 soon
 
I wish Valve would do something to reduce the variance in these survey results. Increase the sample size, discard one-time survey participants from the published findings (that is, only count those who participate, e.g., 2 times in a row), publish results by geography in addition to overall, etc.
 
The survey just says "NVIDIA GeForce RTX 4060". It makes no distinction between the 6GB and 8GB versions, which would be a considerable difference.

Steam probably doesn't include an option for "which variant of the card" in the survey data, because Value never anticipated a company would be sleazy-enough to make a reduced-performance card and not change the name. We're not talking "full-size vs mobile", which is easy to differentiate; it's two "identical" hardware components with disproportionate performance disparities. In any other industry, that would be fraud.

Nvidia should have called the 6GB version "RTX 4060 Lite" or something, but they intentionally used the same name to deceive customers and create market confusion...and then (to go slightly off-topic) the same thing happened again a few days ago. It was just discovered that Nvidia is intentionally disseminating RTX 5080 cards into the wild, through it's board partners, that are missing ROPs. Yet, once again, there is no way to determine which cards those are, until you buy one and check it with GPU-Z.

...and I think "intentionally" is the operative word here. You really think that a company, that purposefully released handicapped products and makes it the customers' problem, accidentally released yet more defective product and told no one? Yeah, no. Press X to "Doubt".

Nvidia is developing a history of deceptive practices and suffers no consequences, seemingly because they are too valuable of a company. Their graphics cards are too important for AI applications, for such "trivial" things as consumer protections, to matter.
 
Last edited:
The survey just says "NVIDIA GeForce RTX 4060". It makes no distinction between the 6GB and 8GB versions, which would be a considerable difference.

Nvidia should have called the 6GB version "RTX 4060 Lite" or something, but they intentionally used the same name to deceive customers and create market confusion.

There is no 6GB 4060, even the mobile version is 8GB. Please find a reference to it anywhere and post that here.
 
I still use an RTX3060 Ti...not much into games, but it did speed up photoshop quite a bit from the built in graphics that was in my last pc
 
I'm happy with my 3060ti and won't be upgrading an overpriced fake frame GPUS.
I'm also "generally" happy with my 3060Ti, HOWEVER I'm running into that VRAM issue with the sheer amount of mods I load for Skyrim, fallout 4 and now Starfield.
With the release of 5000 series from team green I was disheartened about this years upgrade path and was resigned to hanging onto my 3060Ti for another year, but now I'm closely watching team red's releases... If they can keep the price down (even allowing for a 6 month waiting period) I "may" jump onto a 16GB VRAM solution from them...
 
I'm also "generally" happy with my 3060Ti, HOWEVER I'm running into that VRAM issue with the sheer amount of mods I load for Skyrim, fallout 4 and now Starfield.
Then you are one of the few people that should spend the money of a 12GB or 16GB card.
 
I "may" jump onto a 16GB VRAM solution from them...
May? If you're running into VRAM problems, "may" is the wrong word. You need to.

EDIT: Mods, sorry about the double post, I thought it would auto-combine...
 
Last edited:
The survey just says "NVIDIA GeForce RTX 4060". It makes no distinction between the 6GB and 8GB versions, which would be a considerable difference.

Steam probably doesn't include an option for "which variant of the card" in the survey data, because Value never anticipated a company would be sleazy-enough to make a reduced-performance card and not change the name. We're not talking "full-size vs mobile", which is easy to differentiate; it's two "identical" hardware components with disproportionate performance disparities. In any other industry, that would be fraud.

Nvidia should have called the 6GB version "RTX 4060 Lite" or something, but they intentionally used the same name to deceive customers and create market confusion...and then (to go slightly off-topic) the same thing happened again a few days ago. It was just discovered that Nvidia is intentionally disseminating RTX 5080 cards into the wild, through it's board partners, that are missing ROPs. Yet, once again, there is no way to determine which cards those are, until you buy one and check it with GPU-Z.

...and I think "intentionally" is the operative word here. You really think that a company, that purposefully released handicapped products and makes it the customers' problem, accidentally released yet more defective product and told no one? Yeah, no. Press X to "Doubt".

Nvidia is developing a history of deceptive practices and suffers no consequences, seemingly because they are too valuable of a company. Their graphics cards are too important for AI applications, for such "trivial" things as consumer protections, to matter.
I never heard of a 6GB version of the 4060. Did a bit of googling and found no reference to it at all. Even the mobile version has 8GB VRAM. What are you talking about? Didn't you meant the 8GB vs 16GB version of 4060ti??
Can you share some sort of source for a 6GB 4060?? So far you are not making any sense to me.
 
I never heard of a 6GB version of the 4060. Did a bit of googling and found no reference to it at all. Even the mobile version has 8GB VRAM. What are you talking about? Didn't you meant the 8GB vs 16GB version of 4060ti??
Can you share some sort of source for a 6GB 4060?? So far you are not making any sense to me.
So, for reason, I had a "Berenstain Bears" moment.

Turns out, what I remembered being the "4060 6GB vs 8GB models" debacle was actually "Nvidia GeForce RTX 4080 12GB" vs "Nvidia GeForce RTX 4080 16GB" models, and it turns out their VRAM amounts were in the name, so my original criticism was invalid. The 5080 missing parts is still true, though.

Now, you're not going to find those line items in the hardware survey, even if either older 4080 was in there (there is a 4080 Super, but that's just barely a different card), because the branding got changed. Nvidia eventually "unlaunched" the 4080 12 GB and rebranded it as the 4070 Ti.

I apologize for creating confusion. That's my bad.
 
Last edited:
Back