The short version
- GPU: On Valve's default combined-platform table, the RTX 3060 is at 3.62% (+0.08 points). The RTX 5070 is at 3.56% on that table.
- RAM: 16GB is at 41.42% (+3.60 points). 32GB is at 37.06% (−5.16 points).
- OS (all platforms): Windows 11 64-bit is at 72.51% (+1.17). Windows 10 64-bit is at 21.11% (−2.50).
- CPU vendor (Windows only): Intel is at 53.48% and AMD at 46.51%.
Which table are you reading?
Valve publishes several views of the same survey, and the numbers differ between them. Mixing them is how conflicting figures get repeated.
GPU rankings. On the "All Video Cards" view, the RTX 3060 and RTX 5070 are both at 3.78%, with the RTX 3060 listed first. Windowsreport notes that Steam's separate PC GPU table shows the two cards tied at 3.78%. Valve's DirectX 12 GPU view gives yet another pair: 3.99% for the RTX 3060 and 3.93% for the RTX 5070.
Under every view the lead is a rounding-error margin. Guru3D's reading is that the top two are effectively neck and neck in this snapshot. "RTX 3060 leads" is accurate. "RTX 3060 dominates" is not.
One claim to discard. The NoobFeed piece says the RTX 3060 took the top spot back "from its younger sibling, the RTX 3090." Valve's data doesn't support that. The RTX 3090 sits near the bottom of the relevant range, at 0.36% on the combined table. The card the RTX 3060 displaced was the RTX 5070, which led in September.
Operating system. The 72.51% and 21.11% figures come from Valve's combined view, which includes Mac and Linux machines. Valve's Windows-only view shows Windows 11 at 77.38% and Windows 10 at 22.53%. Windows 7 64-bit appears at 0.06%.
Both are correct, but they answer different questions. If you want the share among Steam's Windows machines, 77.38% is the better number. The combined figure is lower mainly because Steam's October OS table shows Windows at 93.70%, Linux at 3.95% and macOS at 2.35%.
Valve's survey doesn't explain why Windows 11 is rising. Windows 10 reaching end of support and new PCs shipping with Windows 11 are plausible contributors, but treat them as context rather than a finding from the survey.
RAM: 16GB is back, but is that real?
On the combined table, 16GB gained 3.60 points. 32GB lost 5.16. September's ranking is reversed again.
| RAM (combined) | October share | Change |
|---|---|---|
| 8GB | 7.52% | +1.05 |
| 12GB | 2.45% | +0.33 |
| 16GB | 41.42% | +3.60 |
| 24GB | 2.31% | −0.25 |
| 32GB | 37.06% | −5.16 |
| 48GB | 1.19% | −0.27 |
| 64GB | 3.94% | +0.23 |
A swing of more than five points in one month is far larger than anything hardware purchases would explain. People didn't pull RAM sticks out of their PCs.
The likely cause is the sample mix. Hardware Busters points out that Valve doesn't weight the survey by region, so a jump or collapse in a market's share of users moves every hardware column. Valve's language table shows a big shift. Simplified Chinese fell to 21.98%, down 9.41 points. English rose to 39.04%, up 4.84 points.
Valve doesn't attribute the RAM and GPU swings to the language change. The language shift is the most visible candidate. As ShiftDelete notes, a big shift in language mix can move hardware numbers without any real change in what gamers own.
Windows-only RAM figures are slightly different: 16GB at 41.27% and 32GB at 38.18%. The ranking is the same.
CPUs: the figures are Windows-only
Valve's Windows-only table has Intel at 53.48% (+0.89) and AMD at 46.51% (−0.90). That is not all-platform processor market share, and it isn't sales data.
For physical CPU counts, the combined table lists eight at 28.68% and six at 27.29%. The Windows-only table shows eight at 28.63% and six at 28.12%. Under the Windows-only view the two are separated by only about half a point, so "six and eight dominate" holds up. "Eight clearly beats six" doesn't. Valve labels these as "CPUs" in its table, and the survey says nothing about what core count is ideal for gaming.
What the survey can't tell you
Participation is voluntary and anonymous. Valve says the data helps it decide on technology investments and products. It is a snapshot of Steam users who took part, not a census of PC gamers.
- It does not measure sales or why people chose their hardware.
- A one-month spike or drop is not a trend. Guru3D makes the point that a one-month spike doesn't show millions of users bought a card, and a return to first place doesn't prove a sales reversal.
- It doesn't show that the RTX 3060 is a good buy today, only that many surveyed systems have one.
NoobFeed's advice that 32GB "should be the default" for new gaming builds is an editorial opinion, not something the survey establishes. There are sensible arguments for it, such as heavy multitasking and background apps. Valve's data doesn't prove them. ShiftDelete also notes that DRAM and NAND have been in short supply this year, making larger kits expensive. That may be the more practical consideration when you build.
Takeaways for Windows users and builders
- Check the table. Quote the RTX 3060 as 3.62% only if you say "combined platforms." Say 3.78% for the PC video-card view, and note it ties with the RTX 5070.
- Windows 11 share depends on the view. The figure is 72.51% across all platforms and 77.38% among Windows systems.
- Don't over-read RAM swings. A 5-point move in a month alongside a 9-point language shift is a sampling story first.
- Developers should keep targeting mid-range systems. For now the typical Steam PC has 16GB of RAM and an 8GB-VRAM GPU. Hardware Busters notes that 8GB cards, at 26.92%, still narrowly lead 16GB cards at 26.35%.
Wait for November's numbers before calling anything a trend. The survey will be more useful once the language mix settles.