Jump to content

NLCbanner2024.jpg.2478be509670e60c2d6efd04834b8b47.jpg

When will shortages in Astro equipment improve?


Recommended Posts

46 minutes ago, Richard_ said:

Of course, it only takes one incident to cause shortages. Back in circa 2012 some fab(s) in Thailand flooded which shut the fab down. This caused a shortage in hard drives / SD cards which is turn ramped the prices up temporarily. We're seeing a similar thing here, but on a global scale.

I remember this as i had just built a gaming PC for myself before this happened. If i recall correctly the prices at least doubled and stayed higher than the before normal pricing for years afterwards.

Link to comment
Share on other sites

19 minutes ago, ONIKKINEN said:

I remember this as i had just built a gaming PC for myself before this happened. If i recall correctly the prices at least doubled and stayed higher than the before normal pricing for years afterwards.

Yeah that's right. Good thing you built it then though, imagine trying to buy a good graphics card today! 

Link to comment
Share on other sites

I am still being quoted 12 months delivery on many semiconductor items.
On many devices, the only stock is from dodgy dealers who provide limited provenance or traceability, and want 5x or 10x the expected price.
I don't see that changing anytime soon.

In part the semiconductor shortage has been caused by manufacturers practices.

If you go back to the early days of integrated circuits (chips) a manufacturer would almost shovel sand (silicon) into one end of a factory and ship completed devices out of the other end.
OK the reality was that different processes were carried out at different factories. But importantly, all under the same ownership with close communication on production matters.
There used to be lots of different manufacturers who offered devices that were often interchangeable directly, or with minor circuit changes.

There has been a gradual takeover, amalgamation, absorption of semiconductor manufacturers, call it what you will. This has reduced the supplier options.
Further, the increasing complexity of the devices has meant an increasing number of components are now single source.

In the last 10/20 years, it has become common practice to send partly finished parts around the world for different parts of the manufacturing process.
Importantly, the factories performing the different processes being under different ownership.
You may have heard the term 'fabless' used in connection with semiconductor companies.
It is not an abbreviation for 'fabulous products'. It means they can't make anything in house because they don't have a manufacturing factory. All manufacture is subcontracted.

Costs have driven more manufacturers to subcontract more parts of their process to specialist companies applying the known price vs volume practices.
Now we have the situation where all the expensive/difficult aspects chip manufacture are done in a very few places on the planet. you can count them on one hand.
If you want to set up to compete. The startup costs are astronomical.

You then ask what happens if there is an earthquake, a tsunami, a contagion, a war, or something else that could knock out these few factories?
With some of these adverse events, we have already seen the consequences. The situation being blamed on covid is in the public eye, but it is not the first.

Putting things bluntly to describe today's situation. A company can make (say) 1 million microprocessors in a year.
They have a guaranteed order from a big car maker who needs engine control units. A customer who has a purchase history, and a future as EVs need even more electronics.
They have a scope mount maker who is looking for 1000 chips. An alarm company looking for 10000 chips, etc. You get the idea.
Who is the microprocessor manufacturer going to keep happy?

Now, if the car makers are struggling, what hope is there for us peasants?

Another consequence of the subcontract culture has been the ease with which counterfeit or 'failed final test' products appear for sale.
But that is a story for another day.

David.

 

  • Like 2
Link to comment
Share on other sites

1 hour ago, Richard_ said:

Yeah that's right. Good thing you built it then though, imagine trying to buy a good graphics card today! 

My GTX1080 proved to be a great value purchase in 2016. I think it will stay in its socket until it bursts into flames 👍.

Link to comment
Share on other sites

Having worked in the semiconductor industry for 35 years, I can explain a few things about chip costs and manufacturing.

Chips were cheap when they were small and contained relatively few transistors.  Practically any company could afford to make them.  However, you could forget about portability or battery power for all but the smallest of chips.

People demanded more functionality in smaller and more power efficient packages.  This necessitated more transistors and larger chips.  This also necessitated shrinking transistor sizes.  This increased manufacturing costs and startup costs.

Chips being more complex required more and more process steps to complete them, requiring longer and longer manufacturing times.  If you cancel an order today, it can take 9 to 12 months or longer to get back in the queue during good times.  Automakers and others foolishly cancelled chip orders early on in the pandemic due to a panic reaction, not fully realizing the consequences of their actions.  Foundries simply filled those manufacturing slots with orders from further in the future that were going to wait for the automaker orders to complete.  PC, laptop, and tablet sales and semiconductor orders skyrocketed, more than making up for the lost automaker orders.  Future slots filled up fast.  The automakers and many others were shutout in the near term.

Fast forward to today, and it costs 7 to 10 billion dollars to build a cutting edge semiconductor foundry to manufacture chips.  Even mighty Intel is struggling to keep up with Samsung and TSMC (the two largest leading edge contract chip houses or foundries).  All the other leading edge legacy chip design houses have sold their legacy foundries and have gone fabless (outsourced their chip manufacturing).  Most legacy fabs in the US are owned by Global Foundries.  These legacy foundries make "good enough" chips for many purposes, but at an elevated price per part at yesterday's performance and power levels.  The US military is one of their biggest customers since they try to source as much from within the US as possible.  One problem for the US military is that there are no microprocessors made in the US anymore.

Unless governments such as China and the US are willing to subsidize new foundries in their own countries, it's unlikely this consolidation of semiconductor manufacturing will change.  You have to be able to sell billions of chips per year to recoup your foundry investment costs.  This requires taking in orders from outside of your own company even for companies like Samsung that sell hundreds of millions of consumer products a year.

Even building a small custom chip requires a $10,000,000 up front cost for manufacturing setup and reservations.  Thus, a lot of companies are moving toward FPGAs for low volume (less than 10,000 units or so) production.  These chips are programmable but cost more per unit on a gate level functionality basis.  It might cost $100 to $2000 for a single FPGA that a custom ASIC in full production volume might cost $5 to $100.  However, if you amortize in the setup costs, they come out cheaper for low volume production.  FPGAs are also larger, hotter, and draw more power than their ASIC equivalents, so they're not a drop-in replacement for all applications.  Also, no FPGA can hold an entire cutting edge microprocessor's worth of gates or run at their elevated clock speeds.  Thus, FPGAs are best for smaller designs and/or coprocessors to offload compute intensive operations from the microprocessor.

  • Like 2
Link to comment
Share on other sites

Its not just chips. It's wood, building materials, steel,  all worldwide.  I'm  looking at ordering a new car and I've discovered theres a shortage of leather for trimming the so many now only come with the bits you sit on in leather and the sides and rears in faux materials.

It's a perfect storm of supply chain inconveniences. Covid affecting both production and supply chains, several massive building projects worldwide sucking up resources - including HS2 in the UK -, significant changes in demand due to different working patterns, and dozens of other factors. The ripples will be felt for a few years yet.

Link to comment
Share on other sites

14 hours ago, Roy Cropper said:

Its not just chips. It's wood, building materials, steel,  all worldwide.  I'm  looking at ordering a new car and I've discovered theres a shortage of leather for trimming the so many now only come with the bits you sit on in leather and the sides and rears in faux materials.

It's a perfect storm of supply chain inconveniences. Covid affecting both production and supply chains, several massive building projects worldwide sucking up resources - including HS2 in the UK -, significant changes in demand due to different working patterns, and dozens of other factors. The ripples will be felt for a few years yet.

Absolutely correct.  I just don't know the details involved in other industries as intimately as I do semiconductors.

Link to comment
Share on other sites

15 hours ago, Roy Cropper said:

Its not just chips. It's wood, building materials, steel,  all worldwide.  I'm  looking at ordering a new car and I've discovered theres a shortage of leather for trimming the so many now only come with the bits you sit on in leather and the sides and rears in faux materials.

It's a perfect storm of supply chain inconveniences. Covid affecting both production and supply chains, several massive building projects worldwide sucking up resources - including HS2 in the UK -, significant changes in demand due to different working patterns, and dozens of other factors. The ripples will be felt for a few years yet.

The same basic antipattern crops up in many different contexts.

For example, IT project managers are usually incentivized on delivery dates above other considerations, so they are pressured to tweak and polish the plans to show the earliest possible dates to the customer. But many things can go wrong in large, complex projects, and if the plans are excessively tight then a single delay can rip through the dependency network and have a disproportionate effect on the final end date.

Most systems featuring a significant dependency network and insufficient slack/contingency at key nodes (stock reserves, in the case of supply chains) will be more efficient in normal operation but risk breaking down completely in exceptional conditions.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.