Wednesday, August 12th 2026

CXMT Surpasses 90% DDR5 Yield, Challenges Industry Giants

Chinese memory giant CXMT has reportedly achieved a significant milestone in DDR5 production, with manufacturing yields now comfortably above 90%. This is comparable to what major manufacturers like Samsung, SK hynix, and Micron are achieving. According to reports, CXMT uses a 17 nm node for DDR5 chip production and has reached a 90% yield for these chips. This is only 2-3% lower than Samsung's yield on its 10 nm class node for DDR5 production, indicating that CXMT's production quality is matching that of the world's largest memory maker. This is a remarkable achievement for a company that excels in cleanroom construction time, taking only 12 months to build its fab's cleanroom compared to the two years others require. As the company plans to end the year with 350,000 DRAM wafers per month, high yields mean there is little waste or leftover.

Chinese MyDrivers also notes that PC OEMs have mixed reactions to adopting CXMT DRAM in their products. For instance, Dell has completely banned the use of CXMT in its machines, while HP only allows CXMT DRAM in its Chinese PCs. Major Taiwanese brands such as ASUS and Acer have increased adoption in some models, as they have more flexibility in supply chain procurement. However, both companies use CXMT memory for PCs sold in mainland China and some emerging markets, meaning the West is still far from receiving PCs with CXMT memory modules. PC OEMs are also cautious about procuring CXMT memory due to concerns about disrupting their relationships with the big three memory makers.
Finally, it is worth noting that all of the 2027 memory production capacity from the big three makers has reportedly been sold out, prompting OEMs to turn to CXMT for additional capacity. However, CXMT is expected to nearly match Micron's output by the end of this year, and there is no clear projection of how much DRAM the Chinese giant will produce next year. It is assumed that most of the available DRAM will be consumed domestically within mainland China, with little expected to be sold in Western markets.
Source: MyDrivers
Add your own comment

41 Comments on CXMT Surpasses 90% DDR5 Yield, Challenges Industry Giants

#26
svan71
aktpu(Michael) Dell has been a eager trump guy for a long time
So ask yourself: if Dell ships a million laptops with Chinese memory and those chips get hit with a patent lawsuit or an IP exclusion order, what happens to Dell then?
Posted on Reply
#27
Fluke
Shrek
  • CXMT: 17 nm
  • Samsung: 10 nm
I wonder how this reflects on chip power consumption
Scaling for DRAM, SRAM and NAND has slowed to such a crawl that it is effectively over.
Testsubject01Most likely: “could be better, but good enough”.
Posted on Reply
#28
outlw
svan71So ask yourself: if Dell ships a million laptops with Chinese memory and those chips get hit with a patent lawsuit or an IP exclusion order, what happens to Dell then?
They... stop shipping machines in affected markets with CXMT RAM?

Not really sure what the argument here is.

Dell using CXMT as a supplier would not make them liable in any patent/IP lawsuits, that would still be CXMT.
And if there are any blocks on imports, they just drop CXMT as a supplier and continue on as usual.
Posted on Reply
#29
_roman_
svan71So ask yourself: if Dell ships a million laptops with Chinese memory and those chips get hit with a patent lawsuit or an IP exclusion order, what happens to Dell then?
Well it sucks when you are a cheap brand which solders everything.

When it is socket, just remove the affected dram and use it somewhere else.

-- more supply may indicate china uses also, but not only, their own dram and the others struggle with the few remaining dram.

afaik germany already got lenovo laptops with chinese dram
Posted on Reply
#30
DavidC1
Shrek
  • CXMT: 17 nm
  • Samsung: 10 nm
I wonder how this reflects on chip power consumption
It's competitive according to one report. It's within 5%.

The bigger hit is to the company, as it costs them more to produce. But it's not 17 vs 10. CXMT is 16nm and Samsung is 10-11nm. Also I believe they'll find an alternative to EUV to scale. China has 10x the IPOs and new graduates compared to USA.
Posted on Reply
#31
Wirko
outlw6669They... stop shipping machines in affected markets with CXMT RAM?

Not really sure what the argument here is.

Dell using CXMT as a supplier would not make them liable in any patent/IP lawsuits, that would still be CXMT.
And if there are any blocks on imports, they just drop CXMT as a supplier and continue on as usual.
I don't think a company has zero liability regarding the legal status of the components it buys. Of course you have less to fear if you're (Michael) Dell, as mentioned above.

And they don't even have to start selling in "affected" markets if non-affected markets, meaning primarily China, are large enough for them.
Posted on Reply
#32
outlw
WirkoI don't think a company has zero liability regarding the legal status of the components it buys.
I would think, that if they knowingly purchase components that are under legal/political sanctions, especially if they are resold in the jurisdictions where the sanctions have originated from, they would be liable.
If, however, there are no legal proceedings, and no import restrictions in the markets where the component is being resold, I am pretty sure they would be fine legally.

That is, unless 'someone' in one of their major markets insists on causing issues for any company that uses Chinese components in any of their offerings worldwide.
But, there are assuredly no loud mouthed politicians like that anywhere, right? :slap:
Posted on Reply
#33
Panther_Seraphin
Memory as it is struggles to scale beyond ~12nm

EUV isnt going to be a massive game changer in the memory beyond possible throughput by reducing the amount of "double patterning" that may be required, but we arent going to see massive density/speed increases from its usage.
Posted on Reply
#34
b1k3rdude
THANATOSNot like I disagree with you that they will sell It as high as possible, but bigger memory supply thanks to CXMT still helps the prices to not increase that much, because difference between supply vs demand is not as big as without CXMT.
Fair, but it will be marginal at best. To the point that you and I as consumers wont actually see a remote difference from our perspective. I am happy to be proven wrong however.
Posted on Reply
#35
boomheadshot8
the cartel is more powerfull than some countries......
one narco cartel is joke vs one nand cartel (plus it's legal)
they've already been caught in 1990' but they do it again
Posted on Reply
#36
rafipc2025
Shrek
  • CXMT: 17 nm
  • Samsung: 10 nm
I wonder how this reflects on chip power consumption
Who cares? How much more it will take 1 more watt?
Come on :D
Posted on Reply
#37
Panther_Seraphin
rafipc2025Who cares? How much more it will take 1 more watt?
Come on :D

Eh....I think it could scale up pretty quick especially at higher speeds

these parts up at 7-8000mhz I bet are quite thirsty in the power dept and then its the heat as well.
Posted on Reply
#38
DavidC1
Panther_Seraphin

Eh....I think it could scale up pretty quick especially at higher speeds

these parts up at 7-8000mhz I bet are quite thirsty in the power dept and then its the heat as well.
Yes, but remember the power increase also comes from increasing the VDD and VDDQ. The most significant increase came from going from 1.25 to 1.35V VDD, which is the main voltage rail for DDR.

Power is Static + Dynamic power. Static, you can't do much about it, but dynamic power is usually dominant under load and it scales square of the increase in voltage and linear to frequency increase. So 10% increase in voltage requires 21% more power. 10% more frequency is 10% more power.

Again, CXMT is actually in line with competitors in power consumption CXMT was using 7.5W while others were in the 7-7.2W range. That's why they are being used in lower power smartphones and laptops as well. If the difference was significant, you wouldn't see much presence in mobile.
Panther_SeraphinMemory as it is struggles to scale beyond ~12nm

EUV isnt going to be a massive game changer in the memory beyond possible throughput by reducing the amount of "double patterning" that may be required, but we arent going to see massive density/speed increases from its usage.
Yea, scaling in DRAM is essentially dead.

But EUV does seem to make some difference in going below 15nm though, as CXMT is using multi-patterning DUV to achieve this. And obviously there's cost savings along with reduced complexity. It probably takes lot of precision and discipline to achieve this. Beyond that I'm curious to see what alternative to EUV the Chinese will come up with. There sure are alternatives as it usually is in every line of work.
Posted on Reply
#39
Panther_Seraphin
DavidC1But EUV does seem to make some difference in going below 15nm though, as CXMT is using multi-patterning DUV to achieve this. And obviously there's cost savings along with reduced complexity. It probably takes lot of precision and discipline to achieve this. Beyond that I'm curious to see what alternative to EUV the Chinese will come up with. There sure are alternatives as it usually is in every line of work.
Below 20nm EUV makes it possible to do most things in a single run where as you are starting to require double patterning at ~20nm and at 10nm your probably at ~4x patterning to get the relevant accuracy/clarity in DUV

While not ideal for production, China having its own internal market makes that more "expensive" processing a null point in normal usage and the AI boom makes it competitive in the rest of the world from a pricing perspective because the market rates are WAY above what production costs would normally set.

China is trying and is supposedly very close to achieving mass market EUV with domestic development/components. Again this will be more important to things like LongArch/Domestic GPUs as it gives them an equal playing field in terms of Cost/NM density etc to western designs.
Posted on Reply
#41
DavidC1
Panther_SeraphinChina is trying and is supposedly very close to achieving mass market EUV with domestic development/components. Again this will be more important to things like LongArch/Domestic GPUs as it gives them an equal playing field in terms of Cost/NM density etc to western designs.
I don't think EUV is the only way to achieve this. It's the one we currently know. There usually are many alternatives before one path in tech is chosen.

Sometimes it's given up, or given up early for other reasons. Intel struggled mightly with 10nm logic process, which are multitudes larger in size than DRAM. China has 10x the graduates and startups compared to US alone. Ideas come from people.
Posted on Reply
Add your own comment
Nov 21st, 2026 20:55 PST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts