Hacker News Reader

Comment by nicoburns | Hacker News Reader

> Are we looking at a future where home computers are replaced by thin clients and all the power lies in subscription services?

The supply chains for high-end chips are brittle enough that it's a very real possibility we end up with a severe supply crunch such that neither clouds nor individual users can access new chips at anything approaching reasonable prices.

TSMC owns 60% of the foundry market. So if China decides to invade Taiwan, that would likely mean ~60% of CPU and GPU manufacturing capacity permanently destroyed at once. That would be "iPhones are no longer for sale this year, and PCs now cost $5000 if you're lucky enough to get ahold of one" kind of territory.

View on HN

> TSMC owns 60% of the foundry market. So if China decides to invade Taiwan, that would likely mean ~60% of CPU and GPU manufacturing capacity permanently destroyed at once.

While it would certainly be devastating, do note that TSMC has fabs in places that aren't Taiwan. So their entire production wouldn't immediately go offline, and presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories.

If China suddenly decides it doesn't want to export electronics, though, then we're all super fucked. After all, what percentage of those TSMC chips flow through China to get mounted onto PCBs or need major supporting components from one of the "Foxconn Cities" in China?

> presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories

There are rumours from seemingly credible sources that Taiwan has the TSMC factories (at least the ones located in Taiwan) rigged with explosives that they intend to trigger in case of invasion by China (as a disincentive against China invading). So China may well not have any say in the matter.

Why would Taiwan destroying its own assets disincentivize annexation goals that have existed long before computer chips were produced there?

Presumably at least in part because China's just as dependent on TSMC as everyone else (at least for the time being). So it's a form of Mutally Assured Destruction, kind of like nuclear weapons. If they actually have to be used, then everyone's in for a bad time, but seeing as nobody wants that, it acts as a disincentive.

This makes no sense. China doesn't want Taiwan because of TSMC, it wanted Taiwan long before TSMC was a major player. The only effect of destroying factories would be to make Taiwan poorer, while China would still get what it wanted.

You're right that this isn't why China want's Taiwan. But the point is that it would also make China poorer. In fact, it would be highly likely to cause a global recession of a magnitude that could threaten the Chinese government due to pressure from it's own citizens.

It actually wouldn't. It would hit many US companies that rely on TSMC's latest nodes like Nvidia and Apple. And TSMC also has its Nanjing fab in China. What this would do is strengthen Samsung and make China even more determined to accelerate SMIC or Huawei's efforts to build a TSMC equivalent. Apple would be more impacted than Huawei, because Huawei isn't using TSMC at all. So in practice, this would hurt the US the most, since it relies on Taiwan based fabs for leading edge nodes.

Not invading Taiwan:

+ China gets to profit from selling phones, computers, etc. to the west

- China doesn't get to own a piece of land

Invading Taiwan:

+ China owns a piece of land

- China can't manufacture anything the west is interested in.

You're arguing about whether it makes sense for China to invade Taiwan. I agree it doesn't right now, but that's not the topic here. The topic is the claim that Taiwan has explosives around TSMC factories, which is ridiculous. So what exactly is your point?

PS. Rozwiń? Nie o tym jest dyskusja "czy Chiny to zrobią", tylko o tym czy Taiwan zrobił to co powtarzane tutaj bezsensowne plotki mówią. To są dwie zupełnie różne sprawy, mimo że mogą się wydawać tożsame.

If Taiwan is being invaded, the annexation is happening. There's no longer any reason to disincentivize annexation. Destroying the fabs is about denying China a major prize.

Destroying the fabs would hurt the West a lot more than China, which is rapidly playing catch up (while US and EU are not).

The other glaring flaw in this pop-geopolitics narrative is that China already has enormous economic leverage over the West, even without the chip supply chain.

> Destroying the fabs would hurt the West a lot more than China, which is rapidly playing catch up (while US and EU are not)

Is that true? My understanding is that Intel while somewhat behind TSMC, is (along with Samsung) still broadly keeping pace. Whereas SMIC while rapidly improving is still playing catch-up.

I doubt it’s something we could know without it happening.

US has intel and some other options, but it would be a colossal issue and adjustment.

China has its well funded, fast progressing Chinese chiplets, but it would be a colossal issue and adjustment.

All we can tea leaf is this: which party has a better history of making large fast industrial adjustments, and which economy is more reliant on cutting edge chips? I think china wins on both personally, so I would give them the edge, gun to head. But it’s an extremely messy process for either.

Did Hong Kong destroy its financial sector to deny China a "major prize"? If someone were going to invade and occupy your country, would you destroy your huge source of revenue so they couldn't claim it as a "major prize"? And then what? Stay poor? I feel like people who repeat this view (something they read somewhere) haven't really analyzed it in a social, economic, historical, and geopolitical context. Because if you do, there's zero logic to it, given the consequences for the 23 million people who would still be living on the island afterward.

Committing to threats/promises "illogically" gives you a better negotiating position.

Acting "illogically" to spite bad behavior leads to less bad behavior.

No one believes it, so it won't strengthen your negotiating position. It's an unconfirmed rumour of unknown origin, and nobody is taking it seriously. And you're missing the historical context, which makes TSMC irrelevant to China's claims.

Those are much better reasons. (Though I don't think the historical context cancels out a risk of losing TSMC. It just means the motivation wouldn't drop as much.)

That question comes up every time this fact is posted, and it could be for the very simple reason to disincentivize annexation to a later date.

> presumably China would still want to keep selling those products and would have an interest in avoiding destroying those factories

It has been hinted by people who might know something that Taiwan has rigged their factories to explode if China invades to ensure China can't get a hold of those factories. I'm not sure if it is true, but it wouldn't be hard to do (the hard part is ensuring the explosives don't go off for other reasons)

ASML can also disable much of the equipment remotely, from Europe. So even if the buildings aren't actually bombed (they likely would be though), someone presses a button a few thousand miles away and most of it gets bricked anyway.

Dunno if I would trust any remote-bricking-capability in an environment where electronic warfare saturates all communication channels.

I agree, I've often wondered if this is more of a dead man's switch where a signal gets transmitted every X minutes or something and if the system doesn't get a signal in some much-longer-but-not-weeks timeframe everything goes kaput.

Not to mention the number of random individuals, with enough access, who might want to sabotage them in those circumstances. And fuck knows what the Trump administration decides to bomb. And the general fog of war. And how delicate everything is.

These supply chains have very non-linear responses. Relatively small fluctuations in demand can have enormous effects on prices and leadtimes.

It's a one-way journey, unless we can adapt quickly enough to a drastic reduction in the general availability of compute power. IMO, our reliance on bloated tech is an existential risk, and reverting to some reasonable baseline needs addressing as urgently as any other current crisis.

I have been waiting for quite some time for some sort of reckoning with our glut for compute resources, but for years I had optimistically assumed this would be due to physical constraints rather than artificial socioeconomic ones. Now is the most advantageous time to be a retrocomputing enthusiast as the definition of "retrocomputing" may seek to expand to engulf the whole category of home computing.

Destroyed? It's far more likely that China will find a diplomatic solution to prevent sabotage of the foundries. Supply capacity will collapse for the US, but only because it will blanket ban imports of TSMC-made products, i.e. a self-imposed supply shock.

I sometimes wonder if this is the real reason for the AI bubble. Major cloud providers are trying to front-load their computing purchases in case geopolitical trouble makes it impossible to get computers. At cloud scale, that's billions of dollars in new investment. So they kick off a new computing paradigm, hype it up, and now their cost of capital for this new investment is significantly lower because they can sell it as getting a piece of future revenues. Basically just a way to make shareholders pay for geopolitical hedging.

Reply on news.ycombinator.com