In the CPU-Z validation database, 32GB RAM configurations account for 37.2% and 64GB for 7.3%, and both are below their year-earlier levels. On November 6, Doc TB, one of CPU-Z's developers, posted that PC RAM capacity had declined year over year for the first time. However, when the year-earlier shares are back-calculated from the published year-over-year figures, the declines are only about 1 percentage point for 32GB and about 1.6 points for 64GB. Meanwhile, 16GB rose by about 3.8 points. The numbers show that the capacity mix of PCs registered in the validation database has changed. They do not directly show that the amount of RAM installed in existing PCs has actually fallen.
16GB share rises while all four categories of 48GB and above decline
The CPU-Z Validator statistics page published by CPUID lists RAM capacity shares under "Last 3 months (year over year)." The most common capacity, 32GB, accounts for 37.2% but fell 2.5% year over year. 16GB, by contrast, rose to 33.0% (+12.9% year over year), narrowing the gap with 32GB.
| RAM capacity | Share | Year over year |
|---|---|---|
| 32GB | 37.2% | -2.5% |
| 16GB | 33.0% | +12.9% |
| 8GB | 11.0% | +5.5% |
| 64GB | 7.3% | -18.0% |
| 24GB | 2.3% | +19.7% |
| 4GB | 2.3% | -26.2% |
| 48GB | 1.7% | -22.5% |
| 128GB | 0.6% | -28.7% |
The share declines are most pronounced among high-capacity configurations. 48GB fell 22.5% year over year, 64GB fell 18.0% and 128GB fell 28.7%. 96GB, which is not in the table, also fell 26.5%. In other words, all four categories of 48GB and above listed in the statistics (48GB, 64GB, 96GB and 128GB) are down from a year earlier. 4GB also dropped sharply, by 26.2%, while 24GB grew 19.7%. Overall, the shares of very small and large capacities are shrinking, while mid-range capacities such as 16GB and 24GB are growing.
The memory standard and CPU data on the same statistics page show related shifts. Among memory standards, DDR4 rose to 54.5% (+6.0% year over year), while DDR5 fell to 32.9% (-2.9%). Among CPU sockets, AM4 rose to 23.5% (+17.5%), and individual CPUs also grew sharply: the Ryzen 5 5500 reached 2.0% (+67.3%) and the Ryzen 5 5600 reached 2.2% (+36.1%). AM4 is a platform that uses DDR4 memory, which suggests the registered PCs are tilting toward the DDR4 generation.
However, the CPU-Z statistics page does not state how the year-over-year change is calculated. For example, 64GB has a 7.3% share and a year-over-year change of -18.0%, which does not mean the share fell by 18 percentage points. It is more natural to read it as a relative change against the share in the same period a year earlier.
Back-calculating the year-earlier shares gives changes of roughly 1 to 4 points
Because the CPU-Z statistics page does not list year-earlier shares themselves, they have to be back-calculated from the current share and the year-over-year figure. Assuming the year-over-year figure is a relative change, the formula is: year-earlier share = current share ÷ (1 + year-over-year change). Since shares total 100%, when one capacity's share rises, the shares of the others necessarily fall in relative terms.
Using this method, the 16GB share rose by about 3.8 points, from roughly 29.2% a year earlier to 33.0%. By contrast, 32GB fell by about 1.0 point, from roughly 38.2% to 37.2%, and 64GB fell by about 1.6 points, from roughly 8.9% to 7.3%. The gap between 32GB and 16GB has also narrowed, from about 8.9 points a year earlier to 4.2 points.
Applying the same calculation to memory standards, DDR4 rose by about 3.1 points, from roughly 51.4% to 54.5%, and DDR5 fell by about 1.0 point, from roughly 33.9% to 32.9%. The AM4 share also increased by about 3.5 points, from roughly 20.0% to 23.5%. It is worth noting that the shares of 16GB, DDR4 and AM4 all rose over the same period.
However, what these figures establish is only that the composition of PCs registered with CPU-Z Validator has changed. Because the data does not track the same PCs from last year to this year, it cannot distinguish between individual PCs having less RAM installed and a change in the types of PCs being registered. It is therefore not appropriate to interpret the year-over-year drop in the 32GB and 64GB shares as a decline in the amount of memory installed in PCs.
The move to Validator V3 may have affected the statistics
In August 2026, CPUID released a new version, Validator V3, of CPU-Z Validator. The company described the aim of the overhaul as keeping the tool's overclocking DNA while making it useful to a broader range of users. Standard validation completes in under 10 seconds via the "Validate" button, checking more than 100 items, including outdated drivers, CPU and SSD overheating, and unnecessary preinstalled software, and then displays the PC's condition. A validation tool that had mainly been used to share overclocking records has become easier to use for general PC diagnostics as well.
For the previous V2 (V2.15 and later), CPUID said it would accept validations only for 60 days after V3 launched, after which it would switch to read-only. The deadline given in the announcement was October 20. CPUID also said it would add features to the V3 statistics page over time, and would enable many of them once the number of validations in the V3 database reached a scale sufficient for statistical processing. The change in RAM shares was observed immediately after this large-scale overhaul.
The problem is that it is unclear which data is used for the year-earlier comparison. The CPUID public materials we could check do not describe how the year-over-year figure is calculated, nor how data from the old V2 and the new V3 is handled. It is particularly unclear how the period spanning October 20, the stated V2 deadline, is treated, and how much the profile of registered users has changed between last year and this year. If V3 increased participation by general users, that alone could change the capacity shares.
That said, statistics independent of CPU-Z show similar shifts. In the Steam Hardware Survey, the share of 16GB in October 2026 was 41.42%, up 3.60 points from the previous month. 32GB, meanwhile, stood at 37.06%, down 5.16 points from the previous month. A separate survey unaffected by CPU-Z Validator V3 thus also shows 16GB increasing and 32GB decreasing.
However, the Steam results also require caution about sample changes. In the same October 2026 survey, the share of users with Simplified Chinese as their language setting was 21.98%, down 9.41 points from the previous month. Because the composition of respondents itself changed significantly, it is difficult to explain the capacity shifts by actual purchasing trends alone. In addition, Steam's 64GB share was 3.94%, up 0.23 points from the previous month, so it does not show the kind of decline seen in CPU-Z. Note also that Steam's comparison is month over month while CPU-Z's is year over year, so the periods compared differ.
Can soaring memory prices explain the change in shares?
Another possible factor is the impact of soaring memory prices. Rising prices may have led PC makers and DIY builders to hold down installed capacity, increasing the share of 16GB models. Memory prices in Japan are indeed high. In a store price survey in Akihabara conducted by AKIBA PC Hotline! from October 17 to 24, the lowest price for 32GB (16GB×2) was 72,980 yen for DDR5-6000 and 32,980 yen for DDR4-3200.
For the same 32GB, DDR5 costs about 2.2 times as much as DDR4. A 64GB (32GB×2) DDR5-6000 kit cost 149,800 yen, about 2.05 times the price of the 32GB kit. Price gaps like these can influence the choice of memory standard and capacity when buying or building a new PC.
A summary of a TrendForce report published in October also projects that contract prices for PC DRAM will continue to rise in the fourth quarter of 2026, though at a slower pace than in the previous quarter. It also says second-tier manufacturers are increasing white-label PCs that use reballed DDR4 memory and older-generation CPUs. As XenoSpectrum reported on November 4, the average spot price of mainstream DDR4 chips (1Gx8, 3200MT/s) fell 0.09% in the week from October 28 to November 3, but buyers remain cautious about purchasing. Against the backdrop of rising prices, the explanation that new PC configurations are tilting toward DDR4 and 16GB has some plausibility.
However, the possibility that memory prices influence capacity choices is a separate question from whether that influence caused the change in CPU-Z's shares. CPU-Z's public statistics have no cross-tabulation of memory capacity and standard, so it is impossible to confirm whether 16GB increased among DDR4 systems or whether high-capacity models are also declining among DDR5 systems. Nor can changes in purchasing behavior caused by price increases be separated from sample changes accompanying the move to Validator V3. On the other hand, because Steam's October survey also shows changes in the same direction, the CPU-Z overhaul alone may not explain everything.
Two points will be important in judging how much the tilt toward DDR4 explains the change. One is whether the 32GB share continues to fall below the previous year even if DDR4's year-over-year change is no longer positive in the next statistics update. The other is whether CPUID publishes capacity shares limited to DDR5 systems and whether the 32GB share is also declining within them. If either is confirmed, it would be difficult to explain the 32GB decline by the tilt toward DDR4 alone.
Conversely, if the 32GB share is maintained among DDR5 systems alone, the view that an increase in DDR4-generation PCs pushed down the overall share would strengthen. Either way, though, what CPU-Z's statistics show remains the composition of PCs registered in the validation database. Determining whether the RAM installed in existing PCs has actually decreased would require data that continuously tracks the configuration of the same PCs. This "first year-over-year decline" suggests that memory capacity choices in the PC market may be shifting, but it must also be read in light of changes in the user base of Validator.