4 ms·

This is far from a complete answer, and not unique to the chip shortage, but some context about some markets and industries in general for those wondering about

by _rend 5y ago

This is far from a complete answer, and not unique to the chip shortage, but some context about some markets and industries in general for those wondering about "how could this be related to COVID?" It's important to understand that many manufacturing industries operate on a "JIT (just-in-time) manufacturing" model[0], where goods used in a manufacturing process are only ordered and received as needed, minimizing inventory and storage costs. (e.g., instead of having a warehouse full of screws sitting around waiting to go into products, you order X screws from another manufacturer every month to fulfill Y number of effective orders) This means that many industries (especially the automotive industry) operate on a constant stream of supplied products, in many interdependent chains. There are two downsides to this approach: 1. The JIT model has to be at least partially predictive. You can't necessarily order components right as you receive orders, because it takes time to produce those components (remember, no one has warehouses full of those components ready to ship, because it's often JIT all the way down) and you don't want to keep customers waiting as components get shipped to you. Instead, you need to at least somewhat predict how many orders you expect to fulfill in order to balance between keeping customers waiting ("shortage") and having excess stock laying around which might never sell (waste) 2. The other main downside is that JIT production is highly interdependent on other industries and can be fragile. If you expect to be able to order X screws this month from your supplier, but some external event causes them to only be able to supply you with Y screws (Y << X), you're shit out of luck. You literally cannot produce your product because the underlying components simply do not exist, which leads to a shortage of your product Both of these downsides come to a head in the presence of some global event that causes uncertainty and inefficiency. If a low-level JIT process depended upon by many industries suffers from an inefficiency (e.g. workers can't come in because they're sick), many higher-level processes start to slow down (e.g., no one has screws to build their products with). This effect is multiplicative across industries, since many manufacturers specialize in specific products and downtime can affect many "downstream" clients. Along with this, such global events can dramatically change customer demands, and when you've only predicted needing to produce X products but you now have Y demand (Y >> X), customers are out of luck because the products literally do not exist to be sold (shortage). Considering that computer chips are now present in almost all major consumer products these days, it doesn't take much uncertainty and inefficiency at the "base" for a domino effect to significantly slow down a lot of industries. [0]: https://en.wikipedia.org/wiki/Lean_manufacturing https://en.wikipedia.org/wiki/Lean_manufacturing

mohanmcgeek 5y ago
Surely this would explain the initial supply chain shock. But it has been two years, why isn't the backlog from the initial shock diminishing. I understand that supply is not elastic but two years should have been enough for the producers to ramp up the production. Do they think that this is a temporary spike and building capacity would be unprofitable?
- Kliment 5y ago
  See my other comment in this thread - it makes no sense to ramp up production because the current demand increase is almost entirely due to lack of market confidence and there's no guarantee it will last long enough for new capacity to go online. That's why ~nobody is building new capacity for the affected nodes.
- _rend 5y ago
  > two years should have been enough for the producers to ramp up the production Along with what Kliment said about not investing in building capacity, I don't think this specific statement is necessarily true. Many of the original factors that caused these setbacks in the first place are still ongoing: everyone is still trying to catch up, and demand is still outstripping supply almost everywhere, for many, many products; worker mobility is still extremely limited (across countries and regions) because of the ongoing pandemic, so it's not trivial to ramp up. Short of investing in that capacity and expanding into new regions, I'm sure it hasn't been easy to actually ramp up production again, and changes in customer demand and patterns are still shaking out. (And even for those companies which have been investing into ramping production, two years can be the low end to start seeing returns on investments: plants take time to build, resources need to be allocated and transported, new employees may need to be trained, etc.)
- bsder 5y ago
  > Do they think that this is a temporary spike and building capacity would be unprofitable? Ayup. A lot of the previous decades of chip manufacturing is all about how superior it is to be a "fabless" company. You don't have those nasty capital expenditures and your upside is basically unlimited--while owning a fab means you have to employ a ton of people and your profit margin is effectively capped. The fab folks know that as soon as this shock is over, owning a fab goes back to being crap. So, you make hay while the sun shines, and you don't even think about doing capital outlays. Even if they could find the equipment and people (not at all an easy thing), there is ZERO incentive for the fabs to expand. To first order, it's actually more profitable for them to maintain this shortage. Nobody is going to pass up a sale just to keep a shortage going, however. Eventually, things will equilibrate. However, I suspect that there's going to be a lot less choice in semiconductors for quite a while afterward. The only short-term thing that I'm interested in tracking is how many companies were hoarding and are going to panic now that inventory tax is coming due. That's going to move some commodity microcontrollers back into the system from the hoarders. It will be interesting to see how much comes back into the system in the next quarter or two.

by _rend 5y ago

mohanmcgeek 5y ago

Surely this would explain the initial supply chain shock. But it has been two years, why isn't the backlog from the initial shock diminishing. I understand that supply is not elastic but two years should have been enough for the producers to ramp up the production. Do they think that this is a temporary spike and building capacity would be unprofitable?

Kliment 5y ago
See my other comment in this thread - it makes no sense to ramp up production because the current demand increase is almost entirely due to lack of market confidence and there's no guarantee it will last long enough for new capacity to go online. That's why ~nobody is building new capacity for the affected nodes.
_rend 5y ago
> two years should have been enough for the producers to ramp up the production Along with what Kliment said about not investing in building capacity, I don't think this specific statement is necessarily true. Many of the original factors that caused these setbacks in the first place are still ongoing: everyone is still trying to catch up, and demand is still outstripping supply almost everywhere, for many, many products; worker mobility is still extremely limited (across countries and regions) because of the ongoing pandemic, so it's not trivial to ramp up. Short of investing in that capacity and expanding into new regions, I'm sure it hasn't been easy to actually ramp up production again, and changes in customer demand and patterns are still shaking out. (And even for those companies which have been investing into ramping production, two years can be the low end to start seeing returns on investments: plants take time to build, resources need to be allocated and transported, new employees may need to be trained, etc.)
bsder 5y ago
> Do they think that this is a temporary spike and building capacity would be unprofitable? Ayup. A lot of the previous decades of chip manufacturing is all about how superior it is to be a "fabless" company. You don't have those nasty capital expenditures and your upside is basically unlimited--while owning a fab means you have to employ a ton of people and your profit margin is effectively capped. The fab folks know that as soon as this shock is over, owning a fab goes back to being crap. So, you make hay while the sun shines, and you don't even think about doing capital outlays. Even if they could find the equipment and people (not at all an easy thing), there is ZERO incentive for the fabs to expand. To first order, it's actually more profitable for them to maintain this shortage. Nobody is going to pass up a sale just to keep a shortage going, however. Eventually, things will equilibrate. However, I suspect that there's going to be a lot less choice in semiconductors for quite a while afterward. The only short-term thing that I'm interested in tracking is how many companies were hoarding and are going to panic now that inventory tax is coming due. That's going to move some commodity microcontrollers back into the system from the hoarders. It will be interesting to see how much comes back into the system in the next quarter or two.