US plans to fully cut off Huawei from chip suppliers | Hacker News Reader

US plans to fully cut off Huawei from chip suppliers

arstechnica.com

39 pointsCHB040308548267 comments

China imports more chips as a country than it imports oil. Chips are what the future struggle between USA and West vs China is going to be. Amazing book about this is Chip Wars by Chris Miller.

Chips rule the modern warfare. Why is Russia attacking so many civilian buildings? Part of it might be terrorism, but I think bigger story is that they got starved of chips and precision rockets. Javelins that helped Ukrainians stop the war at the beginning, each have 200 modern chips inside.

The struggle of having modern chips is only going to increase, with more and more weapons becoming full of AI systems. Will China be able to survive chip choke? It's high on their list to improve their capabilities in production, and I believe one of the major goals before they can afford to invade Taiwan

Most weapons do not use <10nm nodes, or can plausibly not use them. In a military context, these devices are more likely to be in research phases (like supercomputer simulations and machine learning) than built into the weapons themselves.

> Javelins ... each have 200 modern chips inside.

Javelins are based on 1980ties technology, hardly "modern" chips.

That is projection! Only NATO puts "AI" and similar "features" into weapons. Partly to make them more expensive. Russia and other countries use cheap and reliable components (you know joke about washing machines).

It is well visible with air defense. Patriot missiles are so expensive, because there is a single missile type. It suppose to be universal, for all types of targets. It is loaded with all sort of chips to handle all possible causes.

Russian air defense have several types of cheap simple missiles, each specialized for narrow type of target. They do not have to fire multi million $ rocket to destroy cheap drone!

You’re getting downvoted but the Javelins have been pretty underwhelming. The domestic Kamikaze drones Russia is using have proven cheaper, safer and more effective.

They've been pretty overwhelming to the Russians. In this report it states 300 fired, 280 tanks destroyed https://www.ndtv.com/world-news/how-small-ukraine-force-is-k...

Javelins themselves were effective, problem was in logistics and delivery of first batches. Many had defective battery after siting in warehouse for decades. And Ukrainian manual was unreadable, like google translated.

Russia has been doing weekly salvos of precision missiles yet we were told they’d run out last Spring. The reality is we don’t know exactly how many they have or are capable of producing but it’s probably a lot.

And those civilian buildings getting hit are often unfortunately the result missile interception (as was the one which hit the farmers in Poland). In fact, Ukraine recently fired a guy for publicly stating this.

Anyways, Chip Wars is a great book. The trends here are definitely interesting.

Modern weapons mostly use mature / old nodes that are easy to harden, but RU is simply that behind on indigenous microelectronics. PRC with indigenous 14/28nm (well on way) will be fine for weapons, even AI based capabilities, they can short/medium term DUV to 7nm for bespoke batches. PRC inidgenizing complete 14/28nm chain can afford to disrupt TW who will still provide 80-90% of advanced nodes to west. Question then is can western industry survive loss of huge % of advanced nodes that basically brings them down to PRC level of access, if not worst off since almost no country is planning to completely indigenize semi supply chain like PRC.

E: @paisawalla, over comment limit

TLDR is US has leverage during peacetime over sanctions, especially US origin tech. But PRC has leverage during war because most of East Asian semi is within striking range - and wholely indefensible. And current western chip strategy still relies on access to east asian semi. CHIPS act hopes to recapture 5-10% leading edge production in 10 years if all goes well, but even then dependant on East Asian inputs. Incidentally East Asian semi players on territories that US needs base access in TW scenario. TW semi chain maybe write off (which itself maybe existential to western semi), but if JP/SKR assists in US in TW scenario, expect Samsung, Tokyo Electron and other suppliers to be craters too. Really only safe bet for major powers is to complete domestic supply chain and ensure sufficient deterrence from homeland strikes. Without it, probably expect PRC to have only working 14/28nm supply chain, while everyone else scrambles. Regional TW war in medium/long 5/10+ year timeline is actually likely to CLOSE semi gap because only way for PRC to deny west advanced nodes that's already denied to her via sanctions/export controls/lawfare is through war. Also keep in mind US/western industries have disproportionate value capture from semi production and hence more exposed to actual loss of production. Which makes entire PRC invading TW for TSMC that depends on western inputs a bad meme, if anything PRC will want to invade/bomb TW simply to deny semi access to west if she has sufficient semi capability herself. Which doesn't mean leading edge, but enough for most commercial/consumer purposes, while killing leading edge for western high tech. Which is not to say semi is reason PRC will move on TW, but in event of TW war, denying west advanced nodes is going to be high up on incidental strategic goals.

This is an interesting, if terse, analysis. Can you spend a few more words elaborating your thoughts please?

Began the chip wars have~ https://youtu.be/k_zz3239DA0

The difference is that a country cannot produce the oil that it does not have. But a large country with a strong economy will end up being able to produce the chips it needs. It will take some time, but they will.

The US's policy here is to impede and slow down China's growth in order to maintain their dominance for as long as possible. That's it. And it will achieve that to a degree but will also push China to develop their own industry.

I have to agree with them that China being the #1 superpower would be a disaster for the western world though, and we should try to prevent it.

I'm sure the US are doing this just for economical reasons. But the leading superpower always exports their culture as well. See how much we align with the US here in Europe, culturally.

And China with its confucianism where the state is more important than the individual and its tendencies to limit free speech and use extensive surveillance... These things would go down very badly in our culture and we'd tear ourselves apart fighting over it. Some people will be pragmatic and accept it because that's where the money is and others will fight to protect their freedoms. We're barely able to keep the peace with the levels of polarisation as they are.

In my opinion breaking the ties and reverting to the cold war status (with 2 superpowers and their own spheres of influence) would be much preferable to accepting Chinese (CCP) influence over the western world. The question though is how realistic that still is at this point.

Thanks to the globalization that solution is long gone.

How many anti-China folks in western world are living by it, by not buying anything "Made in China"?

Looking at how those 1 euro shops are filled with customers, not many.

And that is only one way how western and asian cultures are tied together in the globalization age.

If you look at the list of fabs https://en.wikipedia.org/wiki/List_of_semiconductor_fabricat... you'll see that of all the modern fabs, about half of them are in China already.

You still got the top 2 in Taiwan and Korea, so they won't be able to produce iPhones or Macbooks. But the rest is there already. Even Samsung has a few fabs in China.

How do you count?

> China imports more chips as a country than it imports oil.

By what metric I wonder? Barrels vs chips? Kilograms? Or dollars?

Not doubting the validity of your statement but it was just a question that came to mind as they are such very different commodities.

I want to bushels. But dollars.

Ahh I see, that makes the most sense indeed.

Of course China does have significant oil reserves itself which might offset this a bit. But on the other hand they also have significant production capability of lower-grade silicon.

Isn't it easy for Russia to order US chips with e.g. China as the middleman?

No, because those at least dual chips are not sold to China. Export Control exists for this very reason.

If you look at say an Iranian drone it is full with US commodity chips. Possible sold directly or indirectly through maybe Turkey, Pakistan, China.

Suddenly a backdoor in US chips sounds like an attractive option.

These chips are too simple and too cheap for that. We are talking about controller chips that are sold for a few dollars, 20 years old designs and produced in the millions.

On the other hand, the Qualcomm special order soc mentioned in the article, I wouldn’t be surprised if there was a back door in that.

> too cheap for that

Maybe subsidized then?

An attractive option, or a reasonable certainty?

No because exporters of these types of things are very wary of getting sanctioned for doing so if they’re caught.

But Russia could just (via intermediaries) start a bunch of companies abroad, and let them order the chips.

Don't forget that exporters still have a strong incentive to sell stuff.

The word “just” is doing a lot of heavy lifting here. If “random internet guy” thought of this, I’m sure the fucking US government thought of it, too and works on preventing it going undetected.

Perhaps "just" sounds too easy, but we're talking about a nation here with endless resources.

Anyway, perhaps these chips can be watermarked, so if they are found in some Russian tech we know which middlemen to blame.

You don’t need to get that complicated. Just demand a list of who they sold to, then who they sold to and so on until the trail goes cold.

If you can’t get an answer, sanction everyone involved and you’ll find they’ll stop selling to unknown customers very quickly. It’s similar to banks and KYC rules.

That was one solution off the top of my head. I’m sure the government has something more sophisticated in place that could be more effective.

Does China have de facto control over chip production in Taiwan? If they do, why ever actually invade? If not, why haven't they invaded already?

China does not have such control. But to your question -

> If they do, why ever actually invade?

The power of the CCP surfs on top of a mythology. When ethnic Chinese people run successful, free, democratic countries it undermines that mythology.

Consider how the CCP have trashed Hong Kong, which had limited democratic freedoms only.

In the west, economic considerations are high on our motives. For the CCP, the central viewpoint is about pursuit and consolidation of political power. Economic considerations are all considered in the light of power priorities.

There is a trap that westerners fall into of regarding live-and-let-live as the natural order of things. It is not, it is a set of values. The CCP come at the world from a different mindset.

Their strategy has been to set up their own foundries, which they have. SMIC is a Chinese-owned fab company with sites on the mainland. Because Taiwan is close and shares a language, they have been able to recruit and train staff who know how to operate a fab. SMIC didn't have a good reputation for quality when I last heard (which was ages ago), but that's just a yield/cost issue so I don't think they would have difficulty scaling it up if they are cut off from the west.

SMIC needs photolithography machines from Dutch company ASML. Sale of further machines has already been blocked by US. When their current machines fail for lack of maintenance and spare parts, they can't make chips anymore.

> If not, why haven't they invaded already?

China has historically lacked the capability to land a sufficient number of troops on Taiwanese shores.

> If not, why haven't they invaded already?

Taiwan's defence plan is basically to make invasion so costly, it's not worth the effort.

Are you saying China does not produce chips? Like, ignore the existence of Espressiv, Loongson, SMIC, 10s other companies?

What is your definition of modern chip for a weapon made in late 1990x?

How can China invade itself?

There is no company named Espressiv. If you meant Espressif, they outsource their manufacturing to TSMC. TSMC is in Taiwan, not China.

Loongson is also fabless, they outsource their manufacturing to STMicroelectronics, a Dutch company.

SMIC is the only company you listed in China with actual manufacturing capacity for chips. Problem is, they need Dutch machines from ASML to operate, and such sales were already blocked by US. Unless they can somehow reverse-engineer and start producing photolithography machines, they are living on borrowed time until their current machines fail for lack of maintenance and spare parts.

If I read correctly, ASML is the only major company that China still depends on? With the right amount of investment, China will shortly have an alternative, you know. Its not irreplaceable, nor does it require reverse eng.

Who said "only"?

   >Problem is, they [Chinese] need Dutch machines from ASML to operate, and such sales were already blocked by US...

Gotta love the free market! US decides who a Dutch comapny can sell to.

This will boost adoption of non US chips by everyone else, and eventually other non US based technologies as well, as other companies take care not to be the next one on the target line.

We are back at the PGP export regulations days.

there really aren't that many players worldwide. The US and allies control chip making and can choke anyone at least for now. It would be economically unsustainable for most of the countries, the only one that can even dream of producing their own chips is China, but still it will take decades to catch up with modern Western chips.

NATOs GDP is 45% of world GDP. Add to that Japan (3.4%) Australia (0.9%) and South Korea (1.9%) and you have half of world GDP. How can you compete with this block technologically?

The size of the GDP is really just a part of it. Indonesia for example has a high gdp ($1 trillion+) but Switzerland or The Netherlands can make tech that they can't, currently, produce. So it's not like it's half gdp, half the power. It's more like 80% power.

I think some regional players (middle-east like UAE and Saudi Arabia) have enjoyed some playing field in this Russia war because of the circumstances. This might make it look that they have "power" but they don't. Once things calm down, they'll probably regret these decisions.

> will take decades to catch up with modern Western chips

IMO decades a little pessmistic. PRC generating as much STEM talent as roughly OECD combined, with semi only elevated to first-level dicipline in 2018, PRC is spitting out 30k IC graduates per year. They're still about 200k short, ~520k/720k out of what IC talent white paper estimated PRC needed for semi industry. I wadger that's comparable to total direct semi industry talent globally (US currently ~110k), which is really effort from a handful of countries (NL,US,JP,DE) in hardware industry that historically got 2nd pick on talent due to how well software pays/braindrains. People like to meme about how semi is making magic thinking sand but it's also niche industry so far pursued by a (relatively) small amount of people in a handful of countries. I wouldn't be surprised if PRC gets within 1-2 nodes of west in 10 years, and start grabbing market share selling cheap. Of course by then US will start banning software captibility with PRC hardware, triggering another round of PRC indigenization that eats at US software.

Most of the demand for chips, and the vast majority of the strategically important demand, is not for chips made with state-of-the-art fabbing. Slightly old tech and IP is more than enough to become chip independent.

>the only one that can even dream of producing their own chips is China, but still it will take decades to catch up with modern Western chips.

The reason behind China pushing so hard for RISC-V should be bloody obvious to anyone paying even remote attention to the recent goings on.

The question is whether the west hands China the keys (yet again) by obliging (yet again) their underhanded attempts.

What does ISA have to do with semiconductors? RISC-V is an open ISA, I would have thought that chip fabbing and ISAs are effectively unrelated (the ISA might determine the layout of the silicon but it's not like x86-64 is drastically different... it's all NANDs, right?)

I don't think the "it's IP" argument would hold up, as historically China & CCP don't really care about IP

x86 and ARM are western technologies, technologies superior to RISC-V that China won't be able to access soon (or even right now), acts of espionage aside, with the semiconductor economic sanctions ramping up.

China pushing RISC-V is because they see the writing on the wall and want /something/ that can replace x86 and ARM at relatively low cost, effort, and time; both domestically (to satisfy their own needs) and globally (to once again conquer the world economically).

My personal hope is RISC-V doesn't obtain success and continues as the irrelevant ISA it is today. The true motives behind wider RISC-V adoption do not align with western interests.

Accessing ARM chips isn't a problem for them, secondly markets like China can survive by focusing on their own market.

Taiwan is a country of 24 million people that started from very little at the end of the Chinese civil war. But it had brains and a state-backed plan.

Many countries could have the capability to produce decent chips.

Taiwan relies on Europe to supply it with EUV.

Building the entire supply chain for semiconductors is daunting at best.

EUV is only required for state of the art chips, and it's the same as for my previous comment: The Netherlands are an even smaller country by population. With a long term, strategic vision, many countries could do the same or better.

Now, China has the mass to definitely lead in anything they 'put their mind to' given time.

I was responding to:

> Many countries could have the capability to produce decent chips.

China might be able to, lets see. "Many countries" definitely not.

If Taiwan or Korea can, nothing prevents many countries from doing the same. Certainly it is a certitude that China, India, the UE also can but it also shows that smaller countries can as well with the right governance and vision.

I feel you may not be reading or following this conversation.

My point is: Taiwan cannot make "decent chips" without the cooperation of the US and EU. China may possibly be able to without the US and EU, but roughly no other country could even try to build "decent chips" (i.e. EUV) without the help of the US and EU.

> I feel you may not be reading or following this conversation.

LOL, OK have a nice day.

Taiwan is an incredible success story. How Asia Works (book) covers this.

TPUs were added to the control list last year

Doesn't that make a conflict in Taiwan more likely? I understand the national interest of the west here but it doesn't look like a good solution if the result is war over taiwan.

Yep but what if that is the idea?

Than it's a bad one

I am struggling to make sense of this; if you go to a German mediamarkt (biggest electronic store here) Huawei still have a sizable presence with nice Intel based laptops and more curious android open source phones. All using American chips. This ban will kill this off.

Why now? Seems to be driven by internal US politics or perhaps a preparation for next meeting between US and Chinese officials (which seems to be going worse and worse on purpose).

Alternatively the US sees a strategic threat from Huawei’s HarmonyOS, which if it gets popular, and it requires good hardware to get popular, will be the first real non-US consumer operating system.

Won't intermediate companies in other countries just buy them and turn around and sell to China?

I dunno, you look at what China has accomplished the last 30 years and I struggle to see this being effective in the mid- to long- term. Economics has a funny way favouring people who are more willing to let the market work.

The US government is sitting more in the passengers seat here than the short term policy power they have suggests. The Chinese will only do badly here if the Chinese Communist Party does something stupid (so, to be fair, there are reasonable odds that they will disembowel their own tech industry in a few years).

China "let the market work"? What on earth am I reading.

Counting second order effects of government spending, the US government probably already controls most of the US economy. The bar for China being a more effective free market is lower than you might think, in practice.

They've achieved remarkable results by focusing on making things the world wants, and upskilling. It could plausibly go either way on the chip thing.

Reply on news.ycombinator.com