Global Chip Shortage Threatens Production of Laptops, Smartphones and More


Pretty sure TSMC’s price hike’s been mentioned earlier in this thread but some interesting discussion of likely consequences from 41:40 - 49:30 on Jupiter Broadcasting Coder Radio
It has always been the case the TSMC has higher prices for the latest-greatest node.
Chip production actually has become more costly each time they improve and that's why today you see only a few players (TSMC, Intel, Samsung) that are going into 7nm and smaller.
While it was possible to go to 14nm with existing technologies the transition to 7nm and better means you need to change your production process a lot and these new processes are very prone to fail which reduces yields (working products).
The equipment and facilities for these fabs are way more expensive compared to older nodes. For example the equipment that exposes the pattern of the chip onto the wafer use to cost around 6 million, the latest EUV equipment is already heading to be 100 million or more.
And because the latest-greatest nodes are a bit niche (only high end equipment needs it) TSMC can charge a premium price for these chips, on top of the overhead they have of low yields and paying for the investment.

So I think if Apple, AMD and Intel (yes also Intel) wouldn't be using the latest-greatest node of TSMC they'll have to be longer on that node to get return of investment and be able to invest and switch to even newer nodes.
All the older nodes become cheaper and have higher yields, which makes it possible to get better and cheaper equipment for the rest of the consumers. Which I think is a good thing.
Sure Apple can get dominant with they own CPU and take market share but they'll probably still be more expensive compared to other products. Apple historically didn't really go with lower prices for newer products, but you never know.

The price hike could probably also related to investments TSMC had to make to improve their current production lines and hire more staff.
So in general I wasn't surprised by the price increase.

Also keep in mind that a lot of chips created for mainstream products (cars, washing-machines, microwaves and what not) don't have to be produced on the latest-greatest production process.
So the price hike by TSMC will mostly impact high end products, like gaming consoles, CPU's, videocards, iphone's/ipad's, etc.
 
It has always been the case the TSMC has higher prices for the latest-greatest node.
Chip production actually has become more costly each time they improve and that's why today you see only a few players (TSMC, Intel, Samsung) that are going into 7nm and smaller.
While it was possible to go to 14nm with existing technologies the transition to 7nm and better means you need to change your production process a lot and these new processes are very prone to fail which reduces yields (working products).
The equipment and facilities for these fabs are way more expensive compared to older nodes. For example the equipment that exposes the pattern of the chip onto the wafer use to cost around 6 million, the latest EUV equipment is already heading to be 100 million or more.
And because the latest-greatest nodes are a bit niche (only high end equipment needs it) TSMC can charge a premium price for these chips, on top of the overhead they have of low yields and paying for the investment.

So I think if Apple, AMD and Intel (yes also Intel) wouldn't be using the latest-greatest node of TSMC they'll have to be longer on that node to get return of investment and be able to invest and switch to even newer nodes.
All the older nodes become cheaper and have higher yields, which makes it possible to get better and cheaper equipment for the rest of the consumers. Which I think is a good thing.
Sure Apple can get dominant with they own CPU and take market share but they'll probably still be more expensive compared to other products. Apple historically didn't really go with lower prices for newer products, but you never know.

The price hike could probably also related to investments TSMC had to make to improve their current production lines and hire more staff.
So in general I wasn't surprised by the price increase.

Also keep in mind that a lot of chips created for mainstream products (cars, washing-machines, microwaves and what not) don't have to be produced on the latest-greatest production process.
So the price hike by TSMC will mostly impact high end products, like gaming consoles, CPU's, videocards, iphone's/ipad's, etc.
This is the sort of insightful comment that makes me really appreciate threads like this on this forum - thank you! I learn so much here
 
I think a lot of these makers really need to consider, do I really need that microcontroller? I'd have thought a truck's EMU and indicators could all still be driven using a Z80 manufactured at some μm (as opposed to the nm scales that modern CPUs are made at). Maybe these modern trucks have media playback capabilities and things like android auto though, I wouldn't know, but I'd hope you could do that on a good old OMAP3 manufactured in the USA. But yeah all those other manufacturers putting 64-bit ARM chips in microwaves and washing machines need their heads looking at.
 
do I really need that microcontroller?
Technology is moving fast these days. Appliances come with patches and extra functionality can only be added if there is a big enough eprom and CPU power to make the new idea happen.


So for example, I think this is not overkill:


Neither is this:
 
  • Wow
Reactions: rSl
Buf. It's only new hardware on a pregnacy test chassis! For a moment I thought it was a real pregnacy test and I feared what the input controls were. Did you have to use the test stick as a joystick or you needed to become pregnant to turn right and become non-pregnant to turn left or what ...
 
That rather looks at the situation in the reverse of how I view it. It seems to me easier to build old larger scale devices because you can do it in more places. For example my local fab over in Wales made 200nm wafers, which is plenty small enough to build a Z80 on, but trust Intel to promise to make as many chips in their latest and greatest at much cost as they like.
 
I like the old chips. With less processor power comes more optimization. So does stability in toolchains and compilers. Expected errors are known and can be circumvented.

If the car industry would go at the speed of new CPU's is like asking a consumer to weekly upgrade their laptops... re-install everything, configure, choose your favorite background, etc.

Similar tensions are in big companies: Developers (you NEED this new functionality, endusers are screaming for it) v/s Administrators (don't touch my system, it's running fine and I want my uptime to stay high)
 
  • Like
Reactions: rSl
I rather think the need to update to fix security holes is more down to systems being connected to the internet, which these days counts as a pretty hostile place for newbies. This isn't the 1990s any more. Now if you're building everything with Z80s you're unlikely to be connecting that to the internet, but you're also not likely to run satellite navigation on a Z80 and all sorts of other modern features.
 
  • Like
Reactions: rSl
There security vulnerabilities are by and large FUD, a coercion technique to get users on the latest code, which has numerous benefits to the software providers.
 
That depends largely whose supplying them. I don't trust MS, google or apple to slip security patches across without bundling them with feature creep, and I keep having to turn off new features every time I update firefox these days, but the linux kernel which is where I care most about security glitches is pretty clear about what's a security hole and what isn't if you look at the mailing list or release notes.
 
I think this was my favourite comment just for the lols:
“Power? What Kind? (+1)
freeze128 16 hours ago

The author of this article isn't very technical. They repeat the word Power, but what they really mean is Electricity. There are many kinds of power: Horsepower, Purchasing power, political power, etc. To make matters worse, they conflate the supply chain problem with this power problem, and now we have a power supply problem!

This article is gibberish.”
 
Nah. 20 years ago we were able to do new things all the time. Now I just want my computer to stop annoying me.

AI has a miserable ROI. But I agree it will completely change the face of humanity one day. Hopefully long after I'm gone.
 
  • Like
Reactions: rSl
Innovation is undesirable. Innovation may bring advantages that are so desirable that counter its intrinsic undesirability (or it may not), but in itself innovation just tilts the balance of power to the seller. The innovator has more information on the sold product than the buyer, so the buyer is more helpless in a market the more innovation there is in the market. Nothing is better just because it's new. It has to prove whether it's better. And the one selling it to you won't explain to you its downsides. Since you're deprived of any experience with it, the seller will fool you easier. People with less experience in any given field doesn't have so much to loose, they can be fooled equally well with new or old things, specially in situations where the existing knowledge is less shared or marketing overcomes the available information for buyers, or the product is specially opaque. This is the reason older consumers and more industrial companies buy more conservative, to leverage their ability to choose. Banking is a specially conservative sector in computing, because they know one thing or two about economics, and have enough power to choose (until recently, when IT giants have grown over them).

The capacity of sellers to fool buyers is not infinite, so there's some tendency for vendors to try to have innovation that results in better products, but it's often easier and more profitable to focus on innovation that reduces costs to the vendor and invalidates buyer knowledge by simply being different to what they've used. That's not always enough to ensure sales, so the innovative sellers try to force the sales with network effects, incompatibilities, lobbying and standards/regulation. This results in real or perceived inutility for older things, and therefore waste and overproduction, so pollution, resources depletion and energy overhead.

Again. This is not an absolute. Sometimes innovation can replace something old that is less efficient or more dangerous, or so. It depends a lot on the particular case. But focusing in innovation as a goal is foolish. Innovation may be a useful mean, but not a worthy goal.
When innovation is useful you'll get good things out of it, so measure those things if you want, don't measure innovation.

I'm not worrying about China becoming more innovative. I'm worrying about concentrated markets, and one country supplying everything to the world. If the west could lose innovation but keep buying less innovative products locally, that wouldn't be necessarily bad. It would possibly be more sustainable and more just (worse for sellers but better for buyers). But Chinese innovation is going to increase chinese market domination. The marrket being dominatd by China is not much better or worse than being dominated by anyone else, but the imbalance is dangerous.
 
Back
Top