China to test 3nm EUV lithography machine that was developed by Huawei

quote:

---

In English please?

---

This was the summary of the video in English



Based on this video, China's 3nm EUV lithography machine will start trial production in the third quarter and mass production in 2026.

The 3nm EUV lithography machine was developed by Huawei by integrating technologies from related domestic units. The equipment is located at Huawei's Songshan Lake base.

China's lithography machine technology route uses LDP high-voltage discharge tin vapor luminescence technology, which is a brand-new technology route and has been patented.

This is different from the LPP laser tinning solution that ASML (ASM) of the Netherlands has been using for nearly 20 years.

The Chinese solution is smaller in size, consumes less energy and is more efficient. Incredible."

As far as I know lithography is a print method.

Posters who don’t understand the issue here would do well to listen to the Acquired podcast episodes on TSMC.

TSMC is already on 3nm and will be on 2nm soon. If this article is true, they won’t be on 3nm until 2026. TSMC might be at 1nm by that point.

Hey remember when China said they had quantum radar that could see through US stealth? But then they had to steal info just to be able to make a knock off Temu version of stealth planes? Pepperidge Farms remembers. But it is nice that I didn't have to looking for Chinese nonsense propaganda updates because John Barron is for sure going to start 25 posts a day about it in this board.

quote:

---

TSMC is already on 3nm and will be on 2nm soon. If this article is true, they won’t be on 3nm until 2026. TSMC might be at 1nm by that point.

---

You make a good point but TSMC won't have 1nm until 2030. The difference between 3nm and 1nm is also going to be incremental compared to the difference of going from 7nm to 3nm chips. With those incremental differences in performance and if this new China 3nm EUV lithography is actually producing higher performance 3nm chips, TSMC might have to change plans in investing 30 billion on a new Fab that won't make business sense anymore if Huawei comes in as a market disruptor with these new chips at a lower cost. Grok did a nice breakdown



"The advantage of a 1nm chip over a 3nm chip would primarily come down to improvements in performance, power efficiency, and transistor density, but the extent of those advantages depends on several factors, including how the industry defines "nm" at that scale and the practical limits of physics and manufacturing.
At the 3nm node—already in production by companies like TSMC as of 2025—you’re looking at incredibly small transistors, with gate lengths and feature sizes that push the boundaries of current lithography tech (like EUV, extreme ultraviolet lithography). A 1nm chip would theoretically shrink those features even further, potentially allowing for:

Higher Transistor Density: More transistors packed into the same area means more computational power or smaller, more efficient chips. Going from 3nm to 1nm could, in theory, increase density by a factor of 5-10x, assuming linear scaling (which it isn’t quite, due to design and thermal constraints).

Better Power Efficiency: Smaller transistors typically require less voltage to switch, reducing power consumption. This is huge for mobile devices and data centers where energy costs dominate. A 1nm chip might cut power use by 20-40% compared to 3nm, based on historical node-to-node trends.

Performance Gains: Faster switching speeds and reduced leakage current could boost clock speeds or parallel processing capability. However, gains here might be less dramatic than in the past—maybe 15-30% over 3nm—since we’re hitting diminishing returns as quantum effects (like tunneling) start to mess with reliability.

That said, there’s a catch: the "nm" label isn’t a literal measurement anymore—it’s more of a marketing term. At 3nm, the actual gate length is already larger than 3 nanometers, and at 1nm, it’s unlikely we’re talking about a true 1nm feature size. Instead, it’s about equivalent performance to what a 1nm node might historically imply. Plus, getting to 1nm would require breakthroughs—like maybe atomic-layer deposition or new materials (e.g., 2D materials like graphene)—because silicon’s limits are looming large.
The real-world advantage might not be as massive as the jump from, say, 7nm to 3nm. Thermal management, quantum interference, and insane manufacturing costs (think $20 billion fabs) could cap the benefits. So, yes, a 1nm chip would likely outperform a 3nm chip in efficiency and power, but the gap might not be revolutionary—more like an incremental win, unless some radical tech (like room-temp superconductors) comes along."

quote:

---

Probably as real as their supersonic missiles

---

Yeah, sounds like wishcasting propaganda.

quote:

---

Hey remember when China said they had quantum radar that could see through US stealth?

---

Don't forget their nuclear battery that they bring up every two or three years.

quote:

---

Huawei will never catch up to them, even if they manage to "steal" a machine

---

They actually are using a different method than ASML and claiming it's more efficient.

"Reports suggest Huawei is trialing this tech at its Dongguan facility in China, with a domestically developed system that uses laser-induced discharge plasma (LDP) instead of the laser-produced plasma (LPP) method employed by ASML"