Hacker News Reader

Comment by maxander | Hacker News Reader

This sort of thing shows how scarily centralized electronics production is. If a disruption at one plant can take out 3.5% of global production, how many of these plants are there? On the order of 30? Imagine a terrorist group decides to bomb them all simultaneously; there goes the production of every device that uses these chips, potentially for years. Are they needed for something critical? How many products do we rely on that all come from the same dozen or so places?

View on HN

I remember when Thailand had heavy floods which spiked the cost of hard drives for years.

Here's the first article I found about that incident, claiming HD prices shot up 50% in one week.

http://www.tomshardware.com/news/seagate-wd-hard-disk-drive-...

A large portion of that increase after about the first 3 months was driven by the manufacturers, not actual capacity constraints. They had an excuse to raise prices so they did. I wouldn't be surprised if there were a lawsuit at some point similar to the monitor and dram collusion lawsuits.

Not sure what you mean "manufacturers, not actual capacity constraints". The beauty of price signals is that they distill all the market information down to a simple metric. It incorporates increased demand, decreased supply, availability of substitute products, predictions about the future and more.

A spike in prices is the best way to ensure that the supply is increased ASAP. The spike is a strong indication that the market wants more and it makes it more likely that suppliers will make decisions to satisfy that demand.

Higher prices allow companies to justify overtime, urgent delivery of supplies, rapid decision making on contracts (i.e. not shopping around) and so on.

Or it was due to collusion in a niche controlled by a duopoly (WD and Seagate).

Also captured in the price signal. The higher price (in your example created via price collusion) is a signal to competitors to enter the market.

Not really, because the price increase sint actually due to real costs on Seagate/wd's side - everyone knows they will just drop prices if a competitor joins that's not part of the collusion. You'd be entering a market with huge startup costs, and once you'd ramped up production prices would drop and your margins would be razor thin.

The hard drive market is widely believed to die with the advent in solid state disks.

With a huge capital cost to get into the market, no investor will put money into building new hard drive factories when it will probably be obsolete in 5 years.

Because you can just hop right into the HDD market? Are we pretending it wouldn't take tens of billions of dollars, 5+ years, and patents nobody is willing to sell or license?

I don’t think the terrorist threat is likely. It would have to be on the scale of 30 simultaneous 9/11 attacks. Anything less would just be a temporary inconvenience, 3.5% for a month is only 0.3% of annual production.

Also, a group capable of 30 simultaneous infrastructure attacks anywhere in the world with the goal of disrupting civilization can easily find far better targets than chip fabs, and various intelligence agencies will have infiltrated such a large international group anyway. The notion is interesting but unrealistic.

Terrorists are a simple-to-state example of the problem, but if that doesn't sound convincing, remember that most (or all?) of these fabs are in a region dominated by a superpower that is moving towards dictatorial authority [1], and additionally, that this region is also (one of the many) moving towards a pattern of extreme weather events [2]. There are many possible events that could seriously disrupt such a concentrated industrial base, and only one would have to happen to cause problems.

[1] https://www.theguardian.com/world/2018/mar/17/china-reappoin... [2] https://phys.org/news/2017-02-china-severe-weather-patterns-...

I think a more realistic (and worrisome) scenario is war brekaing out in South East Asia. Fabs should be clear economic targets for opposing forces. There are a number of conflicts which would likely lead to severe disruption on global semiconductor market.

What thirty targets with similar security as chip fabs do you have in mind? I have a hard time coming up with anything other than nuclear plants for massive damage, but they have somewhat stricter access control, I hope. Oil pipelines are much quicker to repair and most countries have strategic reserves. Refineries maybe?

Liquified natural gas depots. Igniting one of these or a full LNG ship in the river or bay near population centers would kill tens of thousands and the fire could not be extinguished, we would have to let it run its course. Nuclear plants have a bit more security than a chip plant and the logistics of damage by nuclear plant takeover kind of require the authorities to do nothing for a long time.

What about critical infrastructure? Dams which could flood big areas are one example which would be absolutely destructive

I don't think flooding one area is on the same level as killing a large part of microchip production for a couple of years. Almost everything has a chip inside these days.

> various intelligence agencies will have infiltrated such a large international group anyway

But will those organizations get in each other’s way and thus cause us to not stop it in advance. Watch or read The Looming Tower for more context. I’m not sure we really learned our lesson for that one.

well, except 9/11 happened so things can happen that the world's intelligence agencies don't know about. China or the us or russia or a few other countries could just send out missiles in 5 minutes and accomplish this. so just assume idiots or madmen don't get in charge of countries. And it already happened.

Superpowers with huge Arsenal’s could do massive damage. I don’t think this is news to anyone though.

If the Korean War restarts (which is less likely than people think, but still more likely than simultaneous terrorist attacks), the world is likely to lose most South Korean export products for a period.

NAND chips are needed in basically everything electronic now. Phones, tv's, watches, ssds, microwaves, cars, medical devices, and a million other products. They should have become a low margin commodity a long time ago. There is price fixing afoot but the SK president deposed for taking bribes cancelled the investigation into it, and SK let the Samsung CEO out of jail after being convicted of bribing the government so it's going to get worse before it gets better.

Reply on news.ycombinator.com