RTX 3060 Drops From First Place in Latest Steam Hardware Survey

PNY XLR8 GeForce graphics card
(Image credit: Shutterstock)

One of Nvidia's Best graphics cards, the GeForce RTX 3060, has dropped from first place in the latest Steam Hardware Survey, falling to third place in just one month. Well, sort of. There was some very unusual data in last month's Steam Hardware Survey, which caused several Nvidia GPUs to spike extraordinarily high in adoption rates. Things have apparently been corrected, with the affected GPUs returning to normal.

The adoption rates, particularly for the RTX 3060 (desktop and laptop variants), GTX 1060, RTX 2060, and GTX 1650 changed drastically in Steam's Hardware Survey during the month of April (which reports on GPU market share in March). The RTX 3060 (desktop) sat at the top, replacing the GTX 1650 as the most popular GPU among Steam gamers and garnering a whopping 6.31% increase in "market share" in just a single month's time.

The second place position was also shaken up with the RTX 2060 replacing the GTX 1060, garnering an impressive (and highly unlikely) 3.42% uptick in popularity among Steam gamers, with the GTX 1060 falling to third place as a result. But the GTX 1060 also saw an unusually high 2.57% uptick in adoption rates as well. Filling out the rest of the top five GPUs in April were the RTX 3070 and RTX 3060 Ti, which both had outlier increases of 2.52% and 2.14%, respectively. Meanwhile, users with RTX 3060 Laptop GPUs saw a big drop to eighth place, where previously (before April) it was in third place.

This was all quite unusual, as the month-to-month GPU changes typically vary by less than 0.5% in the Steam Hardware Survey. Increases and decreases of several percent obviously indicated something went very wrong with Steam's data collection. With that issue fixed, we're greeted by similarly large changes in the opposite direction for May's results.

Steam Survey Results for May 2023

The Steam Hardware Survey for May (which looks at GPU market share for April) has reshuffled all the affected GPUs back to their normal hierarchy from before April. That means the GTX 1650 is once more in first place, with the GTX 1060 in second. There are a few other changes since we last reported on the Steam Hardware Survey in February.

The RTX 3060 desktop card now sits in third place, with the RTX 3060 Laptop GPU right below it, and the RTX 2060 takes the fifth spot. Before (in February), the RTX 3060 desktop card wasn't even in the top 5 list, with the RTX 3060 laptop GPU sitting in third place, RTX 2060 in fourth, and the GTX 1050 Ti in fifth.

However, those rankings can vary by GPU generation, as Steam lumps certain mobile and desktop GPUs together (e.g. GTX 1650 includes laptops and desktops), while newer models are broken apart (e.g. RTX 3060 and RTX 3060 Laptop GPU). Combining all results for a particular model number (so all RTX 3060 SKUs including desktop, mobile, non-Ti, and Ti parts), the RTX 3060 series is by far the most popular with a market share of 13.44%. Second place goes to the GTX 1650 series at 8.83%, then GTX 1660 series at 6.93%, GTX 2060 series at 6.34%, and RTX 3070 series at 6.12%.

In terms of generational popularity, the RTX graphics cards continue to hold a massive lead over the competition, with the 30-series coming in with 29% market share and the RTX 20-series with 11% market share. Unsurprisingly, RTX 40-series meanwhile sits at just 1.61% total, which itself is double last month's figure. AMD's various GPUs aren't doing nearly as well, with 2.97% among all RX 6000-series cards, 1.42% for all RX 5000-series parts, and 3.38% for the RX 500-series. All Vega-series GPUs (which includes integrated parts) account for 2.03%. AMD's latest RX 7000-series remains MIA, as far as the Steam Hardware Survey is concerned.

Swipe to scroll horizontally
Steam Hardware Survey, Adjusted Data From April 1–30, 2023
DX12 GPUsDECJANFEBMARAPR
NVIDIA GeForce GTX 16506.83%6.88%6.69%4.26%6.74%
NVIDIA GeForce GTX 10606.11%5.70%5.65%8.13%5.35%
NVIDIA GeForce RTX 30604.27%4.04%4.75%11.24%5.09%
NVIDIA GeForce RTX 3060 Laptop GPU4.45%4.92%5.24%3.27%4.93%
NVIDIA GeForce RTX 20605.11%4.79%5.04%8.46%4.85%
NVIDIA GeForce GTX 1050 Ti4.77%4.69%4.39%3.25%4.35%
NVIDIA GeForce RTX 3060 Ti2.85%2.95%3.19%5.33%3.42%
NVIDIA GeForce RTX 30702.99%2.93%3.19%5.73%3.22%
NVIDIA GeForce RTX 30502.43%2.66%2.86%2.08%2.96%
NVIDIA GeForce GTX 1660 SUPER2.85%2.74%2.75%2.48%2.94%
NVIDIA GeForce GTX 1660 Ti2.63%2.61%2.55%1.89%2.46%
NVIDIA GeForce GTX 10502.48%2.47%2.25%1.44%2.20%
AMD Radeon Graphics2.17%2.26%2.17%1.30%2.17%
NVIDIA GeForce RTX 30802.11%2.08%2.16%2.77%2.12%
NVIDIA GeForce GTX 10702.00%1.93%1.84%1.45%1.74%
NVIDIA GeForce RTX 2070 SUPER1.75%1.68%1.69%1.83%1.58%
NVIDIA GeForce GTX 16601.68%1.47%1.56%2.66%1.53%
NVIDIA GeForce RTX 3070 Ti1.37%1.38%1.48%2.14%1.50%
NVIDIA GeForce RTX 3050 Ti Laptop GPU1.24%1.32%1.40%0.86%1.40%
NVIDIA GeForce RTX 2060 SUPER1.27%1.27%1.30%2.06%1.38%

Here's the top 20 GPUs from the past month, according to Steam. We're using adjusted data from the API page, where we sum up the entire column of results and then divide each GPU by the total sum. That's because, for reasons unknown, the columns don't sum to 100% — April's total for DX12 systems is 91.53%, for example. We figure anything that isn't a DX12-capable GPU and thus isn't in the table... shouldn't count!

We've also broken down the totals by GPU family, though the Vega group includes both desktop Vega 64/Vega 56 cards as well as mobile Vega GPUs that are of an entirely different performance class.

Swipe to scroll horizontally
Steam GPU Popularity by Generation / Architecture
GPU FamilyDECJANFEBMARAPR
RTX 40-Series0.00%0.47%0.80%0.79%1.61%
RTX 30-Series25.94%26.96%29.04%36.65%29.19%
RTX 20-Series11.57%10.98%11.36%16.59%10.95%
GTX 16-Series16.21%15.95%15.69%12.62%15.78%
GTX 10-Series17.95%17.27%16.54%16.46%15.93%
GTX 900-Series2.85%2.73%2.56%1.88%2.43%
RX 7000-Series0.00%0.00%0.00%0.00%0.00%
RX 6000-Series2.44%2.67%2.71%1.94%2.97%
RX 5000-Series1.43%1.43%1.41%0.93%1.42%
Vega-Series2.32%2.27%2.07%1.24%2.03%
RX 500-Series3.44%3.47%3.29%2.19%3.38%
RX 400-Series0.50%0.50%0.48%0.30%0.47%
Aaron Klotz
Contributing Writer

Aaron Klotz is a contributing writer for Tom’s Hardware, covering news related to computer hardware such as CPUs, and graphics cards.

With contributions from
  • oofdragon
    I don't trust those pools anyway
    Reply
  • InvalidError
    Last month's stats were really weird, almost as if Chinese crypto-miners dumped their whole inventory into internet cafes and those got surveyed within the same survey interval.
    Reply
  • InvalidError
    oofdragon said:
    I don't trust those pools anyway
    Unless you know of a demonstrably better source, those are the only numbers we've got.
    Reply
  • Admin said:
    Steam's Hardware survey has been fixed, putting the GTX 1650 back in 1st place and GTX 1060 in 2nd.

    Blame Nvidia.

    If the 4090 was $699 like the 1080 Ti was back in the day all those 1080p gamers would be 4K gamers.
    Reply
  • InvalidError
    TravisPNW said:
    Blame Nvidia.

    If the 4090 was $699 like the 1080 Ti was back in the day all those 1080p gamers would be 4K gamers.
    Nah, quite sure there is a large number of people unwilling to pay more than ~$300 for a GPU no matter how great it might be. About 50% of people on the Steam survey use GPUs worth less than $300, which means they either spent less than $300 or cannot be bothered to upgrade to something new at today's prices.

    I personally don't care enough about gaming in general to bother spending much over $200 on a GPU.
    Reply
  • Avro Arrow
    The fact that this even happened just proves how untrustworthy the Steam survey is. We're supposed to believe that a video card that came out ~2 years ago is more popular than cards that have been around for years? Yeah, ok.

    I've tried several times to take the Steam survey and it never works. I've put in a support ticket to Valve and they said "We're aware of the issue and we're working to fix it."

    Meanwhile, I've been sporadically trying to get it to work for over a decade with three different motherboards, five different CPUs and four different video cards.

    With shoddy implementation like that, I can't believe that anyone bothers paying attention to it.
    Reply
  • Avro Arrow
    TravisPNW said:
    Blame Nvidia.

    If the 4090 was $699 like the 1080 Ti was back in the day all those 1080p gamers would be 4K gamers.
    Well, no, not all of them, but certainly far more than there are. Then of course, we could add the fact that if the RTX 4090 was $699, then the RTX 4080 and RX 7900 XTX would also be affordable and that would increase the number of 4K gamers even more.

    The only question would be about how many people had 4K displays to use. I just use my 4K TV but not everyone wants to do that. :giggle:
    Reply
  • gggplaya
    I imagine most of the 1650 people are laptop owners. The 1650 was a popular laptop configuration build during the height of the pandemic and during the mining craze. A lot of kids I know of got laptops during Covid for remote school and these laptops were in the sweet spot of $600-$800. Now they're using them to play games.
    Reply
  • InvalidError
    gggplaya said:
    I imagine most of the 1650 people are laptop owners. The 1650 was a popular laptop configuration build during the height of the pandemic and during the mining craze. A lot of kids I know of got laptops during Covid for remote school and these laptops were in the sweet spot of $600-$800. Now they're using them to play games.
    During covid, the 1650 and other 4GB GPUs were the only things people could readily buy for PCs since you cannot mine ETH on those. Also, the 1650 got a lukewarm initial reception for how little performance it brought at the 1050Ti's price point, which meant plenty of stock floating around until it became the only thing people could get for a reasonable price.

    I'm not surprised at all that the 1650 floated to the top.
    Reply
  • hannibal
    Well Steam survey is rather accurate. Ofcourse there are some movement, because it is survey. Not full data. But after the survey is big enough. It is rather accurate.
    Reply