StatCounter itself has flagged the possibility that some of the new Linux activity could be bot-related, including traffic generated by AI agents and other automated workloads, which would make the jump appear much larger than a shift driven entirely by human users.

Oh, it’s because of AI alright. Because Windows users are tired of it being shoved down their f*ing throats.

(I’m just being sarcastic for the fun of it…)

Definitely a mix of both. I estimate bot pollution is likely growing much faster than natural Linux desktop adoption, though. Simply because LLM based bots are appearing everywhere. I wouldn‘t be surprised if Linux Desktop breaks the 25% mark soon just because of bots.

I do buy the argument that the majority of the shift is due to AI bots / agents. Just looking at the robots.txt I had to implement just to try to keep the majority of bots off my websites is insane (literally 160+ lines of user agents).

But there are some other factors, such as the seemingly steady migration of users from Windows to Linux. I don’t buy the Valve hardware argument as much since the majority of those will be used for gaming on Steam, and not likely to be seen by StatCounter or Cloudflare.

Those are just the bots kind enough to specify a real user agent too 🥲

no one? NOBODY could have seen this coming?

People have been claiming that the current year is the “year of the Linux desktop” because of some improvement to Linux usability or action Microsoft took to offend people for decades. The claim has not come true by any reasonable metric for just as long.

Because Microsoft?

Because bots

Microbots? Microbslopt?

found the person who commented without reading the article

the AI explanation and reasoning is weak. what does “largely disappears” mean without numbers? Cloudflare doesn’t do vibes; give me numbers. in any case “desktop” surely means that the user-agent reflects a desktop browser and not a CLI tool, right? otherwise this number would be closer to 90% than 10%. unless the agents are intentionally spoofing browser user-agents i don’t buy it.

no mention of Steam’s and Framework’s hardware offering that come with Linux? no mention of the clear mass migrations?

here’s an AI explanation: LLMs are helping people who otherwise would struggle move to Linux

Crawlers—even the honest ones—have been spoofing browser User-Agent strings (partially or completely) since the beginning of the commercial Internet. Why would they do this? To dodge bad browser-sniffing scripts, which I still run into from time to time.

sure, but is that implying that 5% of this traffic is bad faith or rogue agents? the tools built into Claude Code and Codex surely have their own user-agent settings.

and why would they spoof Linux instead of Windows?

sure, but is that implying that 5% of this traffic is bad faith or rogue agents?

Given the volume of requests that have been semi-DDOS’ing a lot of servers? 5% is probably an underestimate.

the tools built into Claude Code and Codex surely have their own user-agent settings.

“Surely” is like “assume”—it says more about the speaker than the entity being discussed. They almost certainly don’t use completely distinct strings (no one does—they all start with “Mozilla/5.0”); at most they might have tacked an additional bit on at the tail. And that’s if they’re being aboveboard.

and why would they spoof Linux instead of Windows?

Why not? But actually, this may be due to an incorrect assumption on the part of the stats collectors. Typical strings for bots that admit to being bots don’t specify an OS: “Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; ChatGPT-User/1.0; +https://openai.com/bot” as opposed to, say “Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/134.0.0.0 Safari/537.36 Edg/134.0.0.0” or “Mozilla/5.0 (Android 15; Mobile; SM-G556B/DS; rv:130.0) Gecko/130.0 Firefox/130.0”. If they’re dumping anything without OS markers into the “Linux” bucket, that might do it.

what i mean is that Claude Code wouldn’t report as a Linux desktop browser without some nefarious behavior. what i mean by “surely” is that it would be shitty of Anthropic or OpenAI to do that for no benefit on their part. curl, wget, httpie, xh, whatever client use a specific user-agent that would be a reasonable default to use for any agent without fucking around pretending just to skew metrics.

those header report that they’re bots.

10.61% in June 2026, up from 3.56% in May

Tripling in ONE MONTH.

Yeah, I’m sorry everyone. As much as all of us love Linux and (rightfully) hate Microslop, that’s a jump that simply cannot be explained just by human users.

Data from Cloudflare Radar appears to support that explanation. Cloudflare reportedly recorded a similar rise in Linux traffic, but the spike largely disappeared when its data was filtered to include only traffic identified as human. The comparison is not a direct match for StatCounter’s methodology, but it provides a strong indication that bots and automated systems are responsible for much of the sudden increase.

Thing is, “Cloudflare identified this as a normal browser used by a human” doesn’t mean it actually is a normal browser with a human in the driver’s seat. There are going to be both false positives and false negatives (a browser puppeted by Selenium is still a human-usable browser, after all, and could even be taken over by a human in mid-session; by contrast, a browser whose settings have been chosen by a privacy advocate may make some expected indicators unavailable). So it’s hard to make any useful inferences.

Partially sarcastic, but I did convince 3 people to switch to Linux this year

It was bots.

It’s gotta be bots. 10% of desktop traffic?

planksip.org/10-of-the-population-a-snapshot-of-m…

I’m willing to be most people know somebody that is in that list, and most people don’t know somebody who uses linux. That being said, I bet linux usage is not as outward expressing as a lot of those things. Like, you can tell people own cats because you see them on the porch or in the window, but you can’t glance at most linux users’ houses and know that somebody there uses linux.

Because everything else is enshittifying at an increasing rate. News at 10.

Nah, seriously, I remember a time when Macs were considered ultra-reliable ("it just works"), and now when I look at those of my team developing on Macs… jfc it’s a Soulslike level of hell.

I dunno. I have mixture of Windows, Linux and Mac, and the M chip macs are a fucking beast. Stable, stupid fast, silent, and cool (as in temperature).

I am transitioning all remaining Windows workstations to Proxmox VE based VDIs, but for the people that need a total solution on the road and offline, the mac is hard to beat. Win11 ARM as a VM on Apple silicone runs faster, with less power draw and with significantly less heat than the original installnon raw metal. I’m not sure what dev tools your shop is using for the Macs, but we have had no issues.

Did they start counting Steam Decks?

The default user agent string for the browser that comes with the Steam Deck appears to identify it as Linux, so anybody counting that way without explicitly separating it would count it as Linux.

Because SteamOS is just another arch based Linux distro.

I read on here a couple of days ago that 6% was human traffic and 4% was bots.

I believe what you read was from a Microsoft blog which was variously discounted as being (a) a biased source and (b) based on bad processes for generating counter data.

Yep its not definitive. The only real data from that article was from cloudflare which showed some differing numbers. But thats all. I personally think we will know more once we have all the other trackers out there pull independent numbers.

Articles coming out seem to be what people “feel” like it is. Which isnt real data haha. All we can definitively say is that Linux usage is up from multiple trackers (stat counter/Cloudflare/Steam/etc…) but time will tell how much.

It’s a big jump, for sure. There’s something going on, I just don’t trust any explanation MS gives for it.

Old PCs with Linux, Openclaw, and a handful of LLM models installed, configured by ChatGPT, can be up and running in less than an hour. Once running, people can have them do all sorts of tasks.

I don’t buy the bot narrative, but I also won’t say this is a genuine jump in Linux users. Rather, as someone who has done statical research before, a sudden jump like this feels like a change in how that data was gathered plus a rather high point on natural variance.

For starters, why would bot behavior change in a month? Any serious bot network would already be on Linux, so unless bot filtering got significantly worse in a single month, I’m skeptical on that front. I suppose it’s possible a patch to scrapping plugins happened, or a change in bot sensitivity on the researcher side, or something, but you’d assume someone would point that out.

However, Linux adoption explanations are also iffy. The “Year of the Linux” people point to summer being a factor, or maybe Steam machine release. Maybe someone went viral promoting it and somehow… doubled installs? Seems unlikely, although a small jump can be attribed to these factors I suppose.

But back to the statistics – all samples naturally vary over time and this can be a major factor month to month as statical noise exists. If it drops back to, say, 7 or 8%, that’s probably that factor, although this is much less an issue if the sample is large (which I think it is?). It’s also possible there’s a change in measurement error, such as Unknowns getting parsed correctly this month, though I’m not sure if that can explain most of the chunk coming out from Windows. I think these are the explanations I’m leaning towards, though (e.g. the Linux was there all along!)

All I can say from my experience is a jump like this is unnatural and usually points to variance and/or measurement error. Rather than the actual population having a change, the researcher does something different that changes results. It doesn’t necessarily mean the new results are biased (they might be less biased!), but it also doesn’t mean reality has suddenly changed. After all, we’re talking about a doubling of the population; if 300 million use the Internet in the US, that’s 15 million people changing to Linux in a month.

Many people started taking measures against fingerprinting. Fauxx is a new android app that could cause just this sudden jump (if it had a major users base). But plenty of older browser extensions exist. People are fighting back to regain privacy. I suspect that’s what we’re seeing here.

Edit: AI tools and agents make more sense tbh 🫣😔

This reminds me of a similar situation about a year where Win7 market share suddenly rose from 2% to 9%. If you dug into the data it showed that it was mostly isolated to Singapore and the data also showed a similar increase in incredibly uncommon display resolutions and such. You could also do some math and would come to the conclusion that the numbers only made sense if the total number of devices in Singapore increased tenfold, with everyone running Win7 for some reason.

Long story short, you could explain this in detail to some people and some still wouldn’t accept that this wasn’t the return of Win7.

PS: Here’s a telling chart with data that wasn’t fixed from that time:

gs.statcounter.com/…/singapore/#monthly-202409-20…

PPS: And the same thing can be observed for this current fluke:

gs.statcounter.com/…/north-america

Apparently it’s the year of the 720p Linux desktop.

On track for World Domination™

“bUt nO oNe’S sUrE wHy”

Meanwhile at Microsoft:

“Stop using Windows 10. We lied and made Windows 11. It’s all AI bullshit that we will Cortana in 2 years. Please scan your face and taint to login.”

Microsoft FEELS and ACTS like they have a monopoly.

They do not have a monopoly.

People can use Linux. So they are using Linux. Now up to 10% from 5% at the end of last year.

If Microslop doesn’t start making business decisions based in actual reality, then Linux user share is going to hit 20-30% of NA by early next year at the current growth rate.

I think MS doesn’t care about home users anymore. Azure, Office, Enterprise setups etc. are more profitable. Server core licensing alone is ridiculous; you can license whole departments of personal machines with the price of a datacenter license.

Enterprise is ditching windows overseas and once enough of them do, windows is no longer the common ground buisness need to do anything. It’s an expensive outlier.

Either way, we will see after this month if this is a trend or something of a statistical anomaly.

Theres a lot of articles claiming this and that. The only thing we can definitively say is Linux usage seems up or at least browsers are reporting more Linux across the board. Be careful of people who are saying “its all AI” or “its all algorithms” etc…as its just one month.

I like to see linux being adopted to whatever degree by casual users, but I think it’s worth remembering that part of why linux is the way it is is because it doesn’t need to be obsessed with number go up. Regardless of how much of this is authentic user migration vs bots, it’s clear that there is more momentum and care behind linux, and with that higher likelihood of support from tools/games/hardware.

It just means bots are taking over and we can trust even less of the web.

my ublock blocks that page

It’s AI. The source doesn’t track users but website traffic. It’s Linux-based servers trafficking AI usage.

I’ve reloaded and donated twenty five computers in the last few months. All of them are running linux.

Right… No one is sure why… I am sure that it has nothing to do with being force fed AI and being surveiled by your own computer.

I switched a year or so ago. There was only so much bullshit from MS that i could stomach.

Anecdotal but I switched once Steam worked on linux. That was really the one thing holding me back befoee

Linux is getting better at the same rate windows is getting worse. For many use cases, linux is already better than windows, most people just don’t know it yet.

If they’re tracking web use by PCs they know are not servers, then it’s likely due to a combination of local AI computer web scraping and people using SteamOS devices, which are running Linux. Prosumer local AI users often learn Linux if they don’t already know how to use it, because the software runs more efficiently on it. Anyone with a Steamdeck or Linux distro like Bazzite copying it on the desktop can exit to desktop mode and use it like a normal PC.

It’s ironic. Windows went on AI, not just visibly with Cortana, but under the hood too. For example, they spent a ton developing ONNX, even before “AI” was everywhere, and still push it as a format.

Same with the vendors. AMD NPUs (for instance) were only supported on Windows forever; I think they still are. Same with Intel’s.

…And yet Windows is basically useless for local AI, compared to linux.

Yea thanks I’ll wait to see what the steam hw servey says.