Steam Hardware Survey Shows Big Bump in AMD CPU Popularity
As we head into August, we have the strange pleasure of getting another month’s worth of Steam Hardware Survey data. Last month’s record was great for AMD’s CPUs, but not so good for their GPUs. Meanwhile, it’s starting to look like Linux could overtake MacOS as his second most popular Steam gaming platform in the coming months. No doubt thanks to the Steam Deck.
Red CPUs are rampant
Looking at the CPU landscape first, it looks like AMD has achieved a new record Steam user share. According to the latest data, AMD CPU Share Up 2.22% in July. This is a significant increase in one month.
According to the latest statistics, 33.73% of Steam user PCs are equipped with AMD CPUs, while Intel CPUs make up 66.26%. The battle for Zen 3 vs. Alder Lake is nearing its end. The rivalry between Zen 4 Ryzen and Raptor Lake is poised to win the hearts and minds of PC gamers in the coming months.
Again, keep in mind that Steam Deck uses AMD processors, so it might factor into this equation. Another issue is that Valve has not provided clear details on how the data is collected. There’s a lot of opinion on what’s going on, but the only really valid way to collect stats is a completely random sampling, or sampling every PC that logs into Steam. Neither option seems to be currently used.
Red GPU Green and Envy
For GPU sales data, DX12 API data (opens in new tab) as a reference point, as it provides much more detailed information than the whole video card (opens in new tab) table. Also, as an added bonus, it eliminates really old hardware that is incompatible with DX12.
I took Valve’s data and adjusted it so that the total is 100% (i.e. divided by the sum of monthly percentages for all GPUs in the table). (Perhaps the rest of the GPUs weren’t DX12 capable, which adds up to a range of 89.13% to 92.18%.) I’m wondering why our numbers are slightly different than what was stated in the research. If so, that’s the description. Here’s the breakdown.
March | April | May | June | July | %CHG | |
---|---|---|---|---|---|---|
NVIDIA | 82.08% | 81.56% | 81.99% | 81.49% | 82.78% | 1.29% |
AMD | 13.88% | 14.41% | 14.30% | 14.34% | 14.15% | -0.20% |
intel | 3.25% | 3.23% | 2.87% | 3.31% | 2.29% | -1.02% |
RTX 30 Series | 15.76% | 16.52% | 17.54% | 18.61% | 19.74% | 1.13% |
RX6000 series | 0.67% | 1.30% | 1.45% | 1.52% | 1.81% | 0.29% |
The latest Nvidia advances have pushed back the gains AMD started reaping over the last few months. Here you can see the consistent month-over-month growth of the red team, but the increase is unexciting. In fact, AMD’s progress with his RX 6000 series is marginal compared to his RTX 30 series adoption during the same period.
Nvidia GPUs now make up 5.85x more surveyed PCs than AMD GPUs. Looking at current-gen Ampere and RDNA 2 solutions alone, Nvidia holds a 10.9x advantage over him. Currently, AMD’s most popular RDNA 2 GPU is the RX 6600 XT, making up his 0.4% of all PCs surveyed. In contrast, Nvidia’s most popular Ampere GPU is the RTX 3060 for laptops with 3.59% and his RTX 3060 for desktops is slightly lower at 2.76%.
Overall, current generation AMD and Nvidia GPUs make up 21.55% of all PCs surveyed. Meanwhile, the top three most popular GPUs account for 20.18% in total and are all older models (GTX 1060, GTX 1650 and GTX 1050 Ti). The most popular AMD GPU is still the RX 580 with 1.58%.
Windows 11 user shared slip
Finally, we saw an interesting move in the operating system data. Most interestingly, Windows 11 fell in his July, down 0.11% over the month. At the same time, Windows 10 took a step forward with a pretty convincing 1.91%.
MacOS dropped to 1.74% of user share in the latest results, down 0.71%. Conversely, Linux rose as the OS of choice for 1.23% of users. If this pivot continues, it wouldn’t be surprising to see Linux surpass MacOS, and we think a Steam Deck powered by Steam OS could add some weight to the Linux side.
The Steam Hardware Survey provides interesting data, but the policies and methodologies of the survey are rather vague. Valve does not provide any insight into user selection methods, confidence intervals, or margins of error. I’m not even advocating using random or full sampling, which is necessary for the data to be statistically useful. Still, this is the best publicly available information about hardware PC gamers are running right now, and for better or worse, you should take what’s available.