4 ms·

Global Shortage in Computer Chips 'Reaches Crisis Point'

20 points by optimalsolver 6y ago

tengbretson 6y ago
Aside from Apple and Samsung, who have supply advantages, I wonder if other smart phone makers will even be able to launch new devices this year. Or perhaps they will go the route of the pc component makers and just do paper launches.
grillvogel 6y ago
is this when we finally break free from our computer overlords?
- mikewarot 6y ago
  Without computers, the supply chains fail, and most people starve.
- andresopeth 6y ago
  Hopefully here is where big chip manufacturers "innovate" and find cheaper/better ways of producing these while fulfilling global needs.
  - UncleOxidant 6y ago
    That would take a huge amount of investment in R&D that hasn't been happening... and that R&D could take a long time to produce a cheaper, faster way of fab'ing chips. Given the amount of money involved to develop fabs this would be a government-level investment to the tune of tens of $Billions. I say government-level because there aren't a lot of companies capable of that level of investment. Google, Microsoft and Apple come to mind but are they willing to plow tens of $Billions into what would likely be a decade+ of R&D that might not pay off? Perhaps a consortium of of companies could do this but it's hard to get competitors to cooperate. ("Hey, we all need chips so maybe we should cooperate in developing some foundry fabs?")
  - bsder 6y ago
    Why? If I'm a company with a fab, it behooves me to keep this supply constrained market going as long as possible. Fabless semi companies have been beating up the semi companies with fabs for years. Everybody wanted rid of the capital costs of carrying a fab because it was soooo much more profitable. And the fabless companies didn't want to carry any inventory. If I'm a fabbed company, I'm going to charge you through the nose right now. I'm going to bank back as much as I can, because once this blows over, you fabless jerks are going to go back to beating me up. "Fabless" companies personally chap my hide because they don't do any innovation--they just follow the herd. If you are "fabless", you can't really do any innovation on the semiconductors side. All you can do is churn out the same gravel that everybody else churns out because you are all constrained by the same fab rules and design kits.
    - musicale 6y ago
      > If you are "fabless", you can't really do any innovation on the semiconductors side. All you can do is churn out the same gravel that everybody else churns out because you are all constrained by the same fab rules and design kits. Presumably fabless companies (e.g. Apple atm) focus on architecture, features, and system integration rather than semiconductor innovation. Seems to be working out OK for them. I imagine intel is greatly frustrated that being able to innovate in semiconductors hasn't actually helped them beat AMD either. IBM actually sold its semiconductor business a few years ago. It is disappointing though to watch what appears to be a decline of US semiconductor manufacturing and innovation. Without silicon you have more of a "Software Valley" or maybe Internet [advertising] Valley."
      - bsder 6y ago
        > Seems to be working out OK for them. You pick the outlier and claim it's working out. The vast consolidation in the industry demonstrates that it's NOT working out. Everyone is trying to go to bigger volumes because there's no money in smaller volumes. The ARM microcontroller business is a good example. Everybody is generating the same ARM Cortex-M3/M4 with the same peripherals with the same power consumption and the same prices and ...
kazinator 6y ago
> It is incredible that Samsung sells $56bn of semiconductors to others, and consumes $36bn of them itself, finds it may have to delay the launch of one of its own products,” says Campling. How is it incredible? Semiconductors are not fungible; you can't substitute a different chip because the one you need is unavailable. It's not like substituting margarine for butter in a cookie recipe. Samsung likely makes some high value semiconductors that contain its IP and are profitable; it probably doesn't make every single semiconductor that is needed in various Samsung consumer products and it wouldn't necessarily make sense for it to start doing so during a supply shortage. https://www.samsung.com/semiconductor/about-us/location/manufacturing-centers/ https://www.samsung.com/semiconductor/about-us/location/manu... Look, out of seven of Samsung's manufacturing centers, five do chip fabrication. Of those five, three are in China. If there is some geopolitical problem they will be impacted.
- contingencies 6y ago
  Semiconductors are not fungible; you can't substitute a different chip because the one you need is unavailable Actually in some cases you can. There are nominally package and pin-compatible versions of popular designs from multiple vendors, at least lower end ARM chips from STMicro (which are affected) and older designs like the 8051. That said, they may not be 100% compatible in other aspects, such as clockspeed and electrical properties, and most new devices designed internationally prefer higher-end MCUs.
  - musicale 6y ago
    I assume parent was talking primarily about ASICs rather than standard components.
    - contingencies 6y ago
      That's a strange assumption to make. Not sure I follow the logic.

20 points by optimalsolver 6y ago

Aside from Apple and Samsung, who have supply advantages, I wonder if other smart phone makers will even be able to launch new devices this year. Or perhaps they will go the route of the pc component makers and just do paper launches.

grillvogel 6y ago

is this when we finally break free from our computer overlords?

mikewarot 6y ago
Without computers, the supply chains fail, and most people starve.
andresopeth 6y ago
Hopefully here is where big chip manufacturers "innovate" and find cheaper/better ways of producing these while fulfilling global needs.
- UncleOxidant 6y ago
  That would take a huge amount of investment in R&D that hasn't been happening... and that R&D could take a long time to produce a cheaper, faster way of fab'ing chips. Given the amount of money involved to develop fabs this would be a government-level investment to the tune of tens of $Billions. I say government-level because there aren't a lot of companies capable of that level of investment. Google, Microsoft and Apple come to mind but are they willing to plow tens of $Billions into what would likely be a decade+ of R&D that might not pay off? Perhaps a consortium of of companies could do this but it's hard to get competitors to cooperate. ("Hey, we all need chips so maybe we should cooperate in developing some foundry fabs?")
- bsder 6y ago
  Why? If I'm a company with a fab, it behooves me to keep this supply constrained market going as long as possible. Fabless semi companies have been beating up the semi companies with fabs for years. Everybody wanted rid of the capital costs of carrying a fab because it was soooo much more profitable. And the fabless companies didn't want to carry any inventory. If I'm a fabbed company, I'm going to charge you through the nose right now. I'm going to bank back as much as I can, because once this blows over, you fabless jerks are going to go back to beating me up. "Fabless" companies personally chap my hide because they don't do any innovation--they just follow the herd. If you are "fabless", you can't really do any innovation on the semiconductors side. All you can do is churn out the same gravel that everybody else churns out because you are all constrained by the same fab rules and design kits.
  - musicale 6y ago
    > If you are "fabless", you can't really do any innovation on the semiconductors side. All you can do is churn out the same gravel that everybody else churns out because you are all constrained by the same fab rules and design kits. Presumably fabless companies (e.g. Apple atm) focus on architecture, features, and system integration rather than semiconductor innovation. Seems to be working out OK for them. I imagine intel is greatly frustrated that being able to innovate in semiconductors hasn't actually helped them beat AMD either. IBM actually sold its semiconductor business a few years ago. It is disappointing though to watch what appears to be a decline of US semiconductor manufacturing and innovation. Without silicon you have more of a "Software Valley" or maybe Internet [advertising] Valley."
    - bsder 6y ago
      > Seems to be working out OK for them. You pick the outlier and claim it's working out. The vast consolidation in the industry demonstrates that it's NOT working out. Everyone is trying to go to bigger volumes because there's no money in smaller volumes. The ARM microcontroller business is a good example. Everybody is generating the same ARM Cortex-M3/M4 with the same peripherals with the same power consumption and the same prices and ...

kazinator 6y ago

> It is incredible that Samsung sells $56bn of semiconductors to others, and consumes $36bn of them itself, finds it may have to delay the launch of one of its own products,” says Campling. How is it incredible? Semiconductors are not fungible; you can't substitute a different chip because the one you need is unavailable. It's not like substituting margarine for butter in a cookie recipe. Samsung likely makes some high value semiconductors that contain its IP and are profitable; it probably doesn't make every single semiconductor that is needed in various Samsung consumer products and it wouldn't necessarily make sense for it to start doing so during a supply shortage. https://www.samsung.com/semiconductor/about-us/location/manufacturing-centers/ https://www.samsung.com/semiconductor/about-us/location/manu... Look, out of seven of Samsung's manufacturing centers, five do chip fabrication. Of those five, three are in China. If there is some geopolitical problem they will be impacted.

contingencies 6y ago
Semiconductors are not fungible; you can't substitute a different chip because the one you need is unavailable Actually in some cases you can. There are nominally package and pin-compatible versions of popular designs from multiple vendors, at least lower end ARM chips from STMicro (which are affected) and older designs like the 8051. That said, they may not be 100% compatible in other aspects, such as clockspeed and electrical properties, and most new devices designed internationally prefer higher-end MCUs.
- musicale 6y ago
  I assume parent was talking primarily about ASICs rather than standard components.
  - contingencies 6y ago
    That's a strange assumption to make. Not sure I follow the logic.