What the CPU-Z numbers say
TechRadar, citing Tom's Hardware, reports that CPU-Z's hardware survey found 64GB systems down 18% year over year, with 32GB systems down 2.5%. The CPU-Z Validator statistics page shows the underlying figures:
| Category | Share | Year-over-year change |
|---|---|---|
| DDR4 | 54.5% | +6.0% |
| DDR5 | 32.9% | −2.9% |
| DDR3 | 11.3% | −13.3% |
| 32GB total RAM | 37.2% | −2.5% |
| 16GB | 33.0% | +12.9% |
| 8GB | 11.0% | +5.5% |
| 64GB | 7.3% | −18.0% |
| 24GB | 2.3% | +19.7% |
Two details are easy to miss:
- The 24GB bucket grew fastest, but it is only 2.3% of submissions. A small base makes big percentage swings easy.
- 32GB is still the most common configuration, even after its dip.
Why this isn't a "DDR4 is back" story
CPU-Z Validator counts systems whose owners ran the software and submitted data. It does not count retail sales or shipments. TechRadar says the same: the survey measures computers running the application, not monthly PC sales. DDR4's 54.5% share therefore mostly reflects the large number of older machines still in use.
The data also can't prove that high prices caused the shift. TechRadar suggests buyers are settling for smaller capacities, and that is plausible. A change in submission mix could also come from who runs CPU-Z, how the tool is shared, or which regions submit more. My read is that the pricing explanation is reasonable but unproven.
Shares also move in odd ways, so two caveats apply:
- A share can rise because other groups shrink, not because more machines are being added.
- CPU-Z's own page describes its most popular hardware as a three-month year-over-year view. Readers should treat the figures as a snapshot of that dataset.
The pricing backdrop
The shortage is real, and trackers show it. They use different methods, so treat the figures as indicative:
- Capital and Compute's tracker puts a 32GB DDR5 kit at about $429 to $600 as of November 3, 2026.
- It lists 64GB DDR5 (2x32GB) at around $1,360 and 64GB DDR4 (2x32GB) at $552. Its 64GB DDR5 figure is derived from a single tracker's per-GB median, so treat it with extra caution.
- RamRadar's November 9 snapshot has DDR5 desktop memory averaging $19.72 per gigabyte, against $8.54 for DDR4.
- The spot market tells a more mixed story. Weekly spot pricing showed DDR4 1Gx8 dipping 0.09% to $46.82, and buyers reportedly expect a correction in DDR4.
- Prices for the older standard have still climbed hard. The mainstream DDR4 1Gx8 chip went from $12.18 on Nov 12, 2025 to $46.82 on Nov 3, 2026.
Cheaper retail DDR4 isn't a sign that the industry is returning to it. The same tracker data shows DDR4 chips themselves have become far more expensive, and one tracker says suppliers are winding down DDR4 production. That makes the "hope on the horizon" in the headline hard to find in the pricing data.
What this means for Windows PC owners
DDR4 and DDR5 are not interchangeable. Your CPU and motherboard decide which type you can use. Before spending money on RAM:
- Identify your platform. CPU-Z's Memory and Mainboard tabs show the memory type, and you can also check your motherboard or laptop model's spec sheet.
- Check your board's supported speeds and maximum capacity. Many laptops have soldered RAM and can't be upgraded.
- Match your capacity to your workload. For general Windows 11 use, 16GB is workable. Heavy multitasking, virtual machines and demanding games benefit from 32GB. Few users need 64GB, but those who do, such as developers running several VMs or containers, can't easily trade it away.
- Price the whole platform if you're considering a DDR5 move. A new motherboard and CPU often cost more than the memory itself.
For IT administrators, the practical effect is on refresh planning. Memory costs are tied to AI-driven demand for DRAM, so budgeting for 16GB and 32GB fleet configurations and holding off on blanket 64GB specifications is a defensible response. That is analysis based on general industry reasoning, not a finding from the survey.
Bottom line
CPU-Z's survey shows a real shift in the machines it sees: fewer 64GB systems, more 16GB and 24GB systems, and DDR4 still the majority. It does not show manufacturers going back to DDR4 or buyers across the whole market trimming their RAM. Both explanations are plausible, but this dataset can't confirm either. It's a useful signal, and a poor basis for a buying strategy.