Memory is a very cyclical industry. The Chinese analysis I read concludes Korea memory won because Samsung and SK were willing to take losses and expand production during down turns like 2008, which bankrupted European and Japanese competitors.
The lesson for China is to wait for the inevitable memory downturn (unless you think there is an AI super cycle and this will never happen again) and then go all out with subsidies and out-invest incumbents in a brutal war of financial attrition.
this is actually a great cycle for Chinese memory makers.
We are all well aware of the EUVL restriction and the remarkable progress SMIC/Huawei is making in terms of cramping as much transistors using DUVL.
But lets take a moment to tinker with the fact that, here we are debating with Kirin9030 being 100 or 130 MTr/mm2 and we have TSMC shipping right now chips with a Transistor density of 224 MTr/mm2 and prepping to ship 300+ MTr/mm2 chip by next year.
China has a long catch up to do! Hopefully with EUVL breakthrough in next two years they will (out of nowhere) race ahead just like they are doing in NAND and DRAM and other fields like EV, Photovoltaics etc
so I think it's important to have the correct figures before carrying this discussion further
Based on this TSMC way of calculating transistor density. N5 is around 135m, same as 4LPE. N3E is around 160m and N3P is 180m. We don't get to 200m until N2, which ships next year.
Not really sure what your general point is? I think we are trying to understand where this chip is at right now.
If we look at their progress, the techInsight "7nm" chip came out in July of 2022 IIRC and that was a very small ASIC rather than a full featured phone SoC. TechInsight got a hold of this one in Dec 2025. So essentially in a little over 3 years, they went from a fake 7nm (10% worse than TSMC/Samsung) to a fake 5nm (10% worse), which is a 33% improvement in transistor density and 17.4% shrink in critical feature during those 3+ years. That is a worse cadence than TSMC (which was 2+ years I think from N7 to N5), but the latter have the benefit of transitioning to EUV during the progress. But now TSMC has slowed down to 3 year cadence from N3 to N2 (first iPhone using N3 was 2023 and first one to use N2 is 2026).
So if you want to "catch up" and let's say the rumors of risk production this year or next is true, then they could theoretically get the first phone product using EUV process in 2027. Even in a good scenario, they still need to improve their current process for next year's Huawei phone
