>Doesn't seem much of a threat to TSMC.

That's where you are wrong, TSMC is just the producer, but many many more Company are behind it like:

https://www.asml.com/en (Probably the most important Corp in the industry...but no one knows)

http://www.appliedmaterials.com/

https://www.comet-group.com/

And many more, without those chains TSMC could do just nothing.

EDIT: Oh man i forgot Lam Research!!

And the US 'forbid' asml to sell the technology to china, that means quite allot:

https://www.reuters.com/article/us-asml-holding-usa-china-in...

> https://www.asml.com/en

interesting, i wanted to invest in some chip companies and this gives some good startin gpoints to investigate

I think it's less about overtaking TSMC and more about hedging for a situation where China is completely cut off from chip imports. RIP if have to put out new phones and computers which are worse than the previous year's.

The specific process node isn't important. Once they figure out how to do EUV and other manufacturing techniques that limit their domestic industry's advancement, China will be able to compete at the smaller process nodes too.

Western governments can stop ASML from shipping an EUV machine to a Chinese customer, but they can't stop China from developing their own technology.

Don't forget 12 nm is still the best Intel can fab, it might not be world class currently but it's certainly very capable. Imo it's very threatening, especially if they also keep grabbing talent like they seem to be capable of doing.

Don't forget Intel's nanometers are smaller than TSMC nanometers. TSMC 12nm is 25% lower transistor density that Intel's 14nm

Wait what? I had no idea this was a thing, can you explain?

Ever since (I think) 28nm, companies stopped using nm as an actual gate length and changed it to a marketing size.

Edit: https://en.wikichip.org/wiki/technology_node

There is no standard comparison because of different architectures.

that's crazy, and news to me. So now a unit of length is a marketing term. This is even more misleading than branding the word "real" so you can advertise "contains Real(tm) cheese" on food products.

I think it comes down to what feature length you are measuring. Transistor density seems like a better measure.

Intel has 10nm fabrication, which is equivalent to TSMC’s 7nm

Intel doesn't yet have 10nm https://www.tomshardware.com/news/intel-says-first-10nm-desk... company also revealed that its first 10nm CPUs for the desktop won't come to market until the second half of 2021."

10nm Intel CPUs exist. I have seen them with my own eyes. Reading the tea leaves from public statements and articles like that, the yields seem to be crap right now.

I am typing this reply from a Lenovo Thinkbook 14-IIL, that has a 10nm CPU, also known by its codename Ice Lake. It's not a desktop CPU, it's a mobile CPU, but it's enough to counter your claim that "Intel doesn't yet have 10nm".

Their yields aren't very good at 10nm.

One must start somewhere. That’s iPhone 8 age technology, very usable thing. Especially if it gets the only one available in a trade war.

SMIC is already manufacturing 14nm. The goal is to vertically integrate the entirety of the process, and thus gain independence as well as a future advantage.

Threatening Intel should be good enough? Or do those nanometers not translate? (Genuinely curious)

>Or do those nanometers not translate?

It really Depends:

Smaller means, smaller package, more transistors/sq and less power-consumption. But if you can afford to build a bigger chip to have the same transistor count, and you are ok with the higher power-consumption you have in theory the same performance, sometimes you don't even want 'too' small transistors, an example is radiation-hardening.

It is a threat if the lack of talent stalls development of further nodes.

> Doesn't seem much of a threat to TSMC.

Now. It's still brain drain, and it's to boost their industry. Measures such as this do not exist in a vacuum, more will come.