March Steam survey: Intel Arc finally breaks into the charts as AMD and Linux reach record shares

[midian182]

What just happened? Not for the first time, the Steam survey has returned to some semblance of normality following a bizarre month. That means we have a new top GPU, AMD has reached yet another record CPU share, and Intel Arc has finally broken into the main chart.

February was one of those weird months where the Steam survey saw massive changes from the status quo: the RTX 5070 became the top GPU, AMD lost CPU share, Windows 11 was down 10%, and Chinese became the most common language among participants.

The latest results from March's survey are a lot more familiar. The long-running trend of the RTX 3060, RTX 4060, and RTX 4060 laptop swapping places at the top of the GPU chart continues with the RTX 3060 returning to the top and the other two GPUs directly below it. The previous leader, the RTX 5070, is now fifth.

Most popular GPUs among Steam survey participants during March

Looking at the month's best performers, the RTX 4060 laptop was the leader here. There were also some new entries in the main GPU chart: the RTX 5070 laptop, RTX 5050 laptop, and, four years after launch, Intel's Arc GPUs have finally arrived. Valve lists it as "Intel Arc Graphics," so it's likely a laptop variant – probably the integrated graphics in the Core Ultra series mobile processors.

Best-performing GPUs among Steam survey participants during March

Elsewhere on the survey, AMD is back on track to surpass Intel in the CPU chart. Team Red's share among participants saw a rare decline in February, but it hit a new record of 44.1% in March as Intel dropped to 55.8%.

AMD has long looked on track to overtake Intel in the CPU section this year, but the arrival of the critically acclaimed Core Ultra Plus series could slow the red rise.

Unsurprisingly, Windows 11, which was down 10.4% in February, was up 10.5% in March. Windows 10, meanwhile, crashed 14.8%, leaving just over a quarter of participants still using the older Microsoft OS.

March was a good month for Linux fans. The open-source OS reached a record high of 5.3%, helped by the Steam Deck and its Arch Linux-based SteamOS – Arch Linux was the distro with the largest share (0.34%).

February also saw 32GB suddenly become the most common amount of RAM after it jumped 19%. But its 20.3% fall in March and the 13.5% increase in 16GB users mean the latter is back on top.

Finally, Chinese is no longer the most common language. It was down 31% as English regained the crown.

Permalink to story:

March Steam survey: Intel Arc finally breaks into the charts as AMD and Linux reach record shares

 

[yRaz]

I recently switched from Mint to Kubuntu. I still recommend Mint to anyone looking to try Linux and Bazzite isn't far behind, but Bazzite does some things that don't sit well with me. It is Immutable so if you do run into a problem that needs fixing via command line youre basically SOL. I know people new to Linux won't know how to use the command line, they'll Google a problem and copy and paste commands into it. That approach isn't as straightforward in Bazzite. Thatsaid, being Immutable also keeps people from screwing stuff up and breaking it. It's my opinion that people new to Linux aren't going to be playing with the bootloader's in the command line so I don't know why gaurd rails are necessary.

 

[Beerfloat]

RDNA4 is back on the charts, baby!

One of them, at least:
RX 9070 0.15%

Keep on keeping on AMD.

 

[Thatsdisgusting]

And why you didn’t point in your article how much exactly 5070 lost?

That “little” thingy just completely devaluated all these steam charts to me. It was odd back in march, but now…

P.S especially after how they’ve pointed out, many times, how they’ve refined statistic and gathering month ago.

 

[Vixiy]

Hard to trust these charts after last month's blunder.
Apparently that was due to them counting Chinese users?

So do Chinese users not count anymore? They're gamers too.

Maybe they need to do per country charts instead of a worldwide chart excluding one specific country?

 

[Lew Zealand]

Hard to trust these charts after last month's blunder.
Apparently that was due to them counting Chinese users?

So do Chinese users not count anymore? They're gamers too.

Yes you must be 100% correct that Chinese users do not count anymore. The Steam survey completely agrees with your astute assessment with 22.75% of survey respondents speaking Chinese. Which is exactly zero because they do not count.

Did I get that right?

 

[GTvon]

You have to keep in mind that these numbers do not reflect steam users in general in any way shape or form, this is simply the steam users that share their information. Very few people want their privacy invaded in that way. I never do. It's also the reason most people give for leaving Windows and going to Linux which seems strange because they're the ones sharing their information. So I guess data sharing with corporations isn't the big deal that Linux evangelists try to make it out to be.

 

[GTvon]

So are people pulling out 16 GB of their 32 GB and selling it? If you have 32 GB in your system why not leave it there?

 

[bviktor]

I'll never understand the hype around Arch.

 

[Vixiy]

Yes you must be 100% correct that Chinese users do not count anymore. The Steam survey completely agrees with your astute assessment with 22.75% of survey respondents speaking Chinese. Which is exactly zero because they do not count.

Did I get that right?

People speak Chinese outside of China too.

 

[Lew Zealand]

People speak Chinese outside of China too.

Thank you Captain Obvious. Do you have a point?

 

[Vixiy]

Thank you Captain Obvious. Do you have a point?

The point is I heard they fudged up the previous numbers by including people in china when they don't normally do that.

If that's the case, then I think they should break down statistics by country instead of excluding one.

 

[Lew Zealand]

The point is I heard they fudged up the previous numbers by including people in china when they don't normally do that.

If that's the case, then I think they should break down statistics by country instead of excluding one.

You interpreted wrong. The previous month they included an extraordinarlly large amount of Chinese people in the survey and this month and the previous 11 months they included a representative amount of Chinese people. The normal number of Chinese people in the survey is about 20-25% and:

WOW that roughly matches about the % of Chinese people worldwide. Crazy, eh?

 

[Anton Longshot]

Crazy, eh?

I couldn't help notice, repeatedly, that sarcasm seems to be your hobby.
How's that working out for you being "Techspot Elite"?

 

[Lew Zealand]

I couldn't help notice, repeatedly, that sarcasm seems to be your hobby.
How's that working out for you being "Techspot Elite"?

It works out pretty well and you can try it out for yourself if you want. Sarcasm and the badge thing are independent variables so you can do one without the other if you so choose.

 

[Vixiy]

You interpreted wrong. The previous month they included an extraordinarlly large amount of Chinese people in the survey and this month and the previous 11 months they included a representative amount of Chinese people. The normal number of Chinese people in the survey is about 20-25% and:

WOW that roughly matches about the % of Chinese people worldwide. Crazy, eh?

You could've just explained that in the first place.

It would still be interesting to be able to break things down by country. We now know that Linux is less popular with Chinese users (setting up Linux for asian inputs is hectic, and doesn't work correctly in xwayland apps, so I'm not surprised to see low adoption amongst chinese), but what about other countries/regions?

 

[Hecate91]

Some are weirdly defensive for a completely random survey that repeatedly reports a large influx of Chinese users, or has random spikes of certain hardware.
It really would be nice if Valve would just poll everyone to get some idea of more accurate stats, because Valve already knows your hardware specs and it wouldn't be violating any privacy to report what OS or GPU you use.

 

[Lew Zealand]

You could've just explained that in the first place.

You are correct, I should have. However you should also look at the Steam Survey results yourself (it's linked in this article) so you can be informed about what it says and not make assumptions without reviewing the results.

And while the survey info is problematic from month to month, there are still general trends which stand out, one of which is the Language changes. Sometimes the number of Russian speakers goes up 50% and predictably goes down the next month. Again it's not excluding any one group, instead for one month it's oversampling that group very likely from the country with the most speakers of that language, which then returns to typical the following month.

It would still be interesting to be able to break things down by country. We now know that Linux is less popular with Chinese users (setting up Linux for asian inputs is hectic, and doesn't work correctly in xwayland apps, so I'm not surprised to see low adoption amongst chinese), but what about other countries/regions?

Yes, that is a very good assumption about Linux. And you can also see that Chinese users (it has been assumed these are Chinese gaming cafes*) prefer to use Intel CPUs, Nvidia -60 and -70 series GPUs and 32GB memory and 1440p monitors. And apparently they're still using Windows 10, it will be interesting to see when they move to W11.

It would be nice to get a more detailed breakdown but Steam does this for their own internal reasons and we're "lucky" to even get these simple and often flawed results publicly published.

* I just remembered, have a look at the total drive space, these are overwhelmingly less than 1TB with loads of them less than 500GB, suggesting minimal SSD space you might expect from a cafe. Not sure this is a great assumption though.

 

[Vixiy]

You are correct, I should have. However you should also look at the Steam Survey results yourself (it's linked in this article) so you can be informed about what it says and not make assumptions without reviewing the results.

And while the survey info is problematic from month to month, there are still general trends which stand out, one of which is the Language changes. Sometimes the number of Russian speakers goes up 50% and predictably goes down the next month. Again it's not excluding any one group, instead for one month it's oversampling that group very likely from the country with the most speakers of that language, which then returns to typical the following month.



Yes, that is a very good assumption about Linux. And you can also see that Chinese users (it has been assumed these are Chinese gaming cafes*) prefer to use Intel CPUs, Nvidia -60 and -70 series GPUs and 32GB memory and 1440p monitors. And apparently they're still using Windows 10, it will be interesting to see when they move to W11.

It would be nice to get a more detailed breakdown but Steam does this for their own internal reasons and we're "lucky" to even get these simple and often flawed results publicly published.

* I just remembered, have a look at the total drive space, these are overwhelmingly less than 1TB with loads of them less than 500GB, suggesting minimal SSD space you might expect from a cafe. Not sure this is a great assumption though.

Personally Windows 11 and the enshitification of Windows 10 in the past few years pushed me to Linux, despite fcitx being a hassle. A lot of apps I use have gained native wayland support now whether that being by default or requiring an environmental flag, so not being able to use my Japanese keyboard isn't much of a problem anymore. Only problem software for my keyboard atm is Only Office, since the chromium framework they use has been much slower to add wayland than electron, but the new Collabora Office is good enough to use instead.

 

[Axiarus]

I'll never understand the hype around Arch.

I like my stuff to be bleeding edge up to date.

 

[havok585]

Steam hardware survey is NOTHING.
It's opt-in ! you can refuse to share details when the survey happens, so ALREADY a small sample of "medium" global sample. Steam is not even used in a lot of games. Most have their own launchers or different platforms (ubisoft/ea, etc.)

 

[LordVile95]

Are people forgetting that the hardware survey is a tiny number of systems randomly selected and only includes those who agreed to it? It’s not exactly reliable

 

[Gezzer]

Some are weirdly defensive for a completely random survey that repeatedly reports a large influx of Chinese users, or has random spikes of certain hardware.
It really would be nice if Valve would just poll everyone to get some idea of more accurate stats, because Valve already knows your hardware specs and it wouldn't be violating any privacy to report what OS or GPU you use.

That's my whole problem with the survey. It's a snapshot in time of users that have opted in to letting Valve include their system. But how long do they spend collecting the information and what percentage actually opt in? What about users like myself that have more than one system, or aren't using their system when Valve takes the survey snapshot? At best it's interesting to look at, but meaningful and actually useful data? Yeah... not so much IMHO.

 

[MiqWalsh]

I like my stuff to be bleeding edge up to date.

Well, aren't YOU precious? Not everyone is in that kind of fortunate financial position.....so your statement is kind of irrelevant.

 

[Axiarus]

Well, aren't YOU precious? Not everyone is in that kind of fortunate financial position.....so your statement is kind of irrelevant.

That is true. Good thing Linux is free.