Hacker News Reader

Comment by singhrac | Hacker News Reader

Doesn't seem to worry them too much: https://www.thestreet.com/investing/tsmc-posts-record-q2-pro...

We do seem to be in a secular upswing in semiconductor demand, while TSMC remains the last foundry standing at the 7nm node. It seems like Samsung and Intel alternatives are bit too captive (internally competitive) for a large mobile phone producer. I don't think it's about actual 5G chips themselves, it's about ARM processors and mobile SoCs.

It's doubtful to me that UMC and Global Foundries will catch up, but I do wonder at the long term effects. There are many billions of Yuan to spend on solving this problem. In the past they could get enough of the right people to move to China as they did with OLED.

> In the past they could get enough of the right people to move to China as they did with OLED.

TSMC was founded by a "returnee" as well. Obviously he and his family were rather anti-communist (his father was a ROC official) so he fled from mainland China to Hongkong then the US during the Chinese Civil War. He built a career at TI then returned to China (Taiwan) and founded TSMC when the government was investing in the development of hitech industries. [1]

There are quite a few similar examples on both sides (communists and nationalists). For example, and that's quite a twist of history, the 'father' of the PRC missile and space program was an MIT professor, Caltech professor, co-founder of JPL and was even involved in project Manhattan... Then he returned to mainland China in 1955 [2].

These days there are many more mainland Chinese in the West and Taiwan who work in the semiconductor industry and it's not uncommon for them to return to mainland China to take key positions.

There is no doubt that they'll catch up. The only question is how fast.

[1] https://en.wikipedia.org/wiki/Morris_Chang [2] https://en.wikipedia.org/wiki/Qian_Xuesen

> Then he returned to mainland China in 1955.

It's worth remembering that he did that because he was pushed out, during the Red Scare.

Qian Xuesen was stripped of his security clearance in 1950, as a result of the Red Scare, meaning he couldn't work on his areas of expertise any more. He tried to leave the US, but was arrested on his way out and kept under house arrest for 5 years. That's the thanks he got for helping found the US rocketry program, so when he was released, he founded China's rocketry program.

You already see the same happening under the Yellow peril banner.

The next shoe to drop will be if ASML cuts off Chinese chipmakers. ASML absolutely dominates fab tech.

They already did beginning of the year - https://www.reuters.com/article/us-asml-holding-usa-china-in...

China will figure out <7nm on their own but it will take them 5-10 years.

The US has essentially put on the brakes on Chinese tech dominance for a few years but they can't stop the inevitable.

I just hope the competition will be peaceful. Maybe the next generation of leaders will be better on both sides.

>China will figure out <7nm on their own but it will take them 5-10 years.

Or less: https://www.ft.com/content/6eab0c1c-167f-11ea-9ee4-11f260415...

Gonna be hard without EUV lithography

It's not black magic but knowledge and that's what brain drain means.

That's what I am wondering about. Is ASML the weak point? Would a US, Taiwan, or other company be able to develop and manufacture comparable lithography or EUV machines, or do they essentially have a chokehold on the market for now?

You can't fast-track the technology developed by ASML. We're talking about a company with 4,000 engineers (correct me if I'm wrong btw) who have been working on lithography magic for decades. EUV took decades of dedicated research to finally become a reality just in the past few years. They literally vaporize tin into a plasma in order to create the EUV light, which then has to pass through incredibly complex optics on its way to the mask. 97% of the light is lost on its way to the target... This thing consumes megawatts. It's literally magic.

What's odd to me is that - why is it just one group in the Netherlands doing this? vs. interest in other academic or industrial groups?

Vs. say FinFETs or other groundbreaking tech - I remember Intel coming out with the paper however many decades ago and fast forward a few years later, TSMC and Samsung came out with their own products utilizing it.

Or the Dutch government bows to pressure from the US and makes the decision for them.

Or maybe they did it by themselves because they are capable of independent decision making. Just the other day someone here was loudly proclaiming we shouldn't be interested in US politics from overseas because it doesn't affect us. HN should make up its mind about this in a well reasoned way.

They already have.

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

But you see, money can’t buy love.

Yeah, exactly. TSMC has the premier process right now and is at capacity making chips for Apple, Qualcomm and other high margin (i.e. well-paying) customers. Cutting off the oddball folks for PR reasons costs them nothing at the moment.

I’m not sure Huawei is a small value oddball anymore.

Huawei is bigger than Apple. It's not exactly an oddball.

Not as a TSMC fab customer it isn't, not remotely. The overwhelming bulk of Huawei's revenue remains in device manufacturing. They're a very recent entrant into the SoC market, and not moving a lot of product there.

Huawei needs more chips than Apple, and are moving to produce all of their own chips. As a TSMC fab customer it's at least of the same scale as Apple.

Repeating the same lie in a reply doesn’t make it true. Recently there’s been an influx in the past few days of tons of sinophiles.

Is it true or not that Huawei sells more devices than Apple? : https://www.forbes.com/sites/bensin/2020/12/30/huawei-still-...

Is it true or not that Huawei wanted to use HiSilicon for all of their phones? That is obviously true.

The consequence is that Huawei needs more chips than Apple, that would have been sourced from TSMC. Therefore, they had the intention to be a customer of at least equal size as Apple to TSMC.

How is any of this a lie? I fail to see how much money Huawei makes from this or that changes that Huawei needs more processors than Apple.

> Is it true or not that Huawei wanted to use HiSilicon for all of their phones? That is obviously true.

"Wanted to" is not the same as "does". If TSMC wanted to preserve a potential growth market for their fab, they'd prioritize Huawei/HiSilicon. But they don't, because they're at capacity making chips for western designers already. So they can kill off the Huawei business without worry, because it's not costing them anything. Free PR.

They may need more chips but they don't necessarily need the same mix of chips, Huawei devices don't require A13's.

I mean the z80 was still manufactured until recently and used all over in devices, tsmc isn't going to make those on a 7nm prpcess.

And they will have to buy chips from Qualcomm or Mediatek at their low end. Which is still business from TSMC.

So in the context of TSMC, it is irrelevant.

The same goes for their 5G RAN equipment.

So what you are saying is that A who buys less from them is a bigger customer than B who buys more? Huawei buys more SoCs than apple hence they are a bigger customer. So that would make Apple small fry in your world. Not very likely.

Except as just stated in the above comment, Huawei does not actually buy more SoCs than Apple.

By what measure? They have less than half the revenue, and only 9b in profits last year.

Total phone sales, in number of units, which is what matters when figuring out how many chips to make, as well as market share as a proportion of units.

You are also going to need to factor in upcoming Mac sales as well.

And for all we know Apple may move to a multi-CPU strategy for its higher end products meaning you will get significantly more volume.

For all we know Huawei might do so too, plus HiSilicon makes their own modems while Apple buys them from Intel.

Mac sales are essentially insignificant compared to phone sales, and nowhere near enough to make up the gap in units between Huawei and Apple.

Reply on news.ycombinator.com