Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/hindsight-is-2020-when-did-the-semiconductor-shortage-start.15279/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Hindsight is 2020: When did the semiconductor shortage start?

jms_embedded

Active member
Any idea when in 2020 it was widely suspected that the semiconductor industry was entering a shortage?

I've been looking around published media articles and these are the earliest I've found that give hints:
December 2020 is full of many articles from auto manufacturers about the chip shortage, and once you get into Jan/Feb 2021 then the coverage is much more prominent.

My next step is digging into semi-manufacturer earnings calls and see what I can read from the tea leaves there...

...but I'd greatly appreciate it if any of you know any leads to look into.
 
Last edited:
The shortage on 200mm, and 300mm are two very different creatures.

The sudden resurgence of 200mm, and 200mm shortage been there for close to 5 years.
 
Bitcoin dropped to 5K in Mar 2020 and has been climbing since then. This drove the graphics card shortage as miners bought all the cards (and still do). I recall building a PC in Dec 2020 without a graphics card, based on the prices, so it was pretty bad by Dec 2020.

The shortage of 200mm tools has existed for decades. I worked at a 200mm fab in 2008, with plans to expand. The semiconductor equipment manufacturers would only sell new 300mm tools back then, not 200mm. Then in 2011 and 2012 the semiconductor equipment manufacturers refused to support 450mm tools. So it’s this ongoing Cold War between EQP and fab, which EQP has won.
 
Thanks for doing this, it will be certainly very insightful and important for industry plus market to know.
My view on this is that to call it ‘chip shortage’ is an exaggeration on many counts.
First of all there are many types of chips (IC’s) and certainly not all are is short supply, case in point Memory chips are by no means is mfg. Capacity constrained. Somehow this has not been well reported by media and understood by market.
Secondly as one of the other member pointed out in above comment- chips manufactured on 200mm (8 inch) wafers has seen much of the shortages. This issue had been in making for last 5 to 10 years - with many factors playing together, causing a “perfect storm” !
Let me try to point out few factors of this storm:
(i) technology factor - beyond 40nm all fabs went to 300mm but many large applications IC’s like audio, power, rf etc. didn’t need these advanced technologies and continue to be on 200mm wafers. (ii) almost all foundry’s and IDM’s stopped building new capacity for 200mm to support this large and growing application of audio, power, analog, rf etc. (iii) OEM’s like AMAT, Lam and ASML stopped mfg 8 inch tools completely! (iv) many 8inch fabs & foundry struggled financially and shut down in last 10 years - without much of these applications IC’s (analog/ rf/mems) moving into advanced technologies to take advantage of 300mm capacities. (v) and then came Trump admin. ban on chips mfg. at SMiC etc. in China- a large manufacturer of 8 inch wafers. for fabless companies (vi) finally, auto industry supply dept. shutting down in early 2020 pulling back on their orders thinking auto industry will be down under ‘new normal’ - miscalculation based on economist predictions!
So such factors caused a ‘perfect storm’ for mainly - power, audio, rf types of chips manufactured on certain (mature) technologies on 200mm wafers!
Ofcourse the domino effect has caused shortages for other technologies & chips to be affected.
So, it may feel like this shortage happened in 2021 but had been in play for few years if not a decade.
 
So, it may feel like this shortage happened in 2021 but had been in play for few years if not a decade.
I agree there were clear warnings of increases in 200mm wafer demand since at least 2014 --- examples: Design & Reuse quoting an IC Insights report, and Christian Dieseldorff of SEMI has given presentations on 200mm capacity many times in the last 7 years or so.

But that's not what my question is about. I'm trying to find out when in 2020 that it was widely suspected that the semiconductor industry was entering a shortage. (Or call it a "capacity constrained environment" if you like.) The difference is between there being a problem and someone admitting there is a problem.

I've checked a bunch of quarterly earnings call transcripts from TSMC and some of the automotive IC manufacturers (NXP, ST, Infineon, etc.) and almost all of them presented in late Jan / early Feb 2021 for the quarter ending December 2020 mention the chip shortage. The ones presented in late Oct / early Nov 2020 for the quarter ending September 2020 don't; most mention the recovery in the automotive sector, and there are a few cryptic signs that something is going on, but nobody openly saying "Hey, we have a capacity problem coming."

The first major announcement I could find was the Reuters articles published on Dec 4 2020. (There are several articles; in addition to the Continental AG article I mention in my original post, there's a more general one about Conti / VW / Infineon / NXP: https://www.reuters.com/article/vol...production-vw-and-suppliers-say-idUSKBN28E1IN)

I've edited the post with a few more from late Dec 2020. The first mainstream media article about widespread capacity issues seems to be a Dec 17 article from Reuters.

Dec 18 WSJ article mentions:

Industry leaders and analysts have been warning about the bottlenecks for weeks, though they hadn’t caused production stoppages until now.

Jan 8 AP article mentions:

“There have been warning signs about his for months,” said Kristin Dziczek, vice president of industry at the Center for Automotive Research, an industry think tank.

But I haven't found any public mentions of any warnings, prior to December 2020, aside from the general 200mm capacity issue that was gently raised in the preceding years in industry trade conferences/magazines.
 
Last edited:
Bitcoin dropped to 5K in Mar 2020 and has been climbing since then. This drove the graphics card shortage as miners bought all the cards (and still do). I recall building a PC in Dec 2020 without a graphics card, based on the prices, so it was pretty bad by Dec 2020.

The shortage of 200mm tools has existed for decades. I worked at a 200mm fab in 2008, with plans to expand. The semiconductor equipment manufacturers would only sell new 300mm tools back then, not 200mm. Then in 2011 and 2012 the semiconductor equipment manufacturers refused to support 450mm tools. So it’s this ongoing Cold War between EQP and fab, which EQP has won.
I think most underestimate the impact that crypto is having on the semiconductor space. The current ethereum hashrate is around 950Th/s, this is up from about 160Th/s at the end of 2019. Let's call it an increase of 800Th/s to keep the numbers round.

A top of the line GPU is about 40Mh/s. Do the math and that means 20 million top of the line GPUs diverted to Ethereum mining in the past 2 years. A top of the line GPU is about $600. That's about $12b in GPU sales that have gone to ethereum mining alone in the last 2 years. AMD revenue over the last 2 years is roughly $26b and NVidia is about $28b.

This little back of the envelop math implies that up around 20% of AMD and Nvidia's combined sales (and likely a much higher percentage of high end stand alone GPU sales) over the past 2 years have been diverted to Ethereum mining. That's pretty substantial.
 
Interesting to note, Tesla delivered a record 936,172 vehicles in 2021, as outlined in a report from Tesla, representing an 87 percent increase over the 499,550 vehicle deliveries Tesla made in 2020. So no real shortages there. The other auto companies still seem to be struggling. It will be interesting to see what the investor calls report this month. COVID is a good CEO excuse for lack of planning and execution so we will see some or that for sure. My bet is that labor issues will be the biggest challenge in 2022 due to the COVID mutations and quarantines.

According to Gartner semiconductor revenue growth in 2019 was -12%, 2020 was +7.5%, and 2021 is expected to be 20%+. TSMC revenues will increase 20%+ this year as well. Most semiconductor companies will report big 2021 revenue increases except Intel. 2022 is any ones guess but it will not be as big as 2021. Analysts are saying 2022 semiconductor revenue will grow from 4%-14%. My guess is another double digit year (10%+) which is still much higher than the norm.

So the question really is: When will consumer spending (demand) get more in line with manufacturing (supply)?

Personally, we have been spending more on durable goods and less on entertainment due to COVID. In 2019 we froze expenses completely due to the fear, uncertainty, and doubt of COVID. In 2020 we started spending again, did some home improvements, and in 2021 we over spent, mostly electronics and we experienced no shipping delays. In 2022 our spending will be closer to 2020, our new normal per say.

Home improvements are running rampant in our neighborhood and most projects have been delayed due to the labor and material shortages. We will skip home improvements for another year in hopes things will catch up. Entertainment will still be light but we do plan on a vacation or two. We only had one vacation last year (one week) and none in 2019. We own a sailboat and have some upgrades planned, mostly electronics but installation has been delayed due to COVID quarantines. Even routine boat maintenance has been delayed due to the Omni surge.

So who really knows what is going to happen with COVID. The labor issues are the real problem now and that will hit the supply chain pretty hard in the coming weeks. We had high hopes this time last year due to the vaccine then the delta variant hit. Now it's Omni so your guess is as good as mine.
 
The authorities seem to be optimistic with covid and normalcy returning for the most part. A large risk was taken with the vaccine and it was worked well beyond expectations.

Omicron and flurona could of been a lot worse.
 
I think we will have some level of normalization by spring. By then we should have updated boosters, not to mention a fairly significant chunk of the population will have already recovered from Omicron. We'll get some reprive over the summer, but who knows what new variant we'll be facing before the end of the year. Hopefully there is enough population level immunity that it doesn't cause serious concern.

The general pattern is that with each successive wave of Covid, we encounter something more infectious, but the case outcomes are generally improving thanks to better immunity and treatments. If this trend continues Covid will become something we deal with like the flu.
 
The general pattern is that with each successive wave of Covid, we encounter something more infectious, but the case outcomes are generally improving thanks to better immunity and treatments. If this trend continues Covid will become something we deal with like the flu.

I certainly hope that this is true. I thought we would get a handle on this after the first wave with the vaccine. Then came Delta and now Omni. Who's to say that an even more deadly version isn't in the wings?

I like the flu narrative, but again, I would not bet the life of my grandchildren on it.
 
The authorities seem to be optimistic with covid and normalcy returning for the most part. A large risk was taken with the vaccine and it was worked well beyond expectations.

Omicron and flurona could of been a lot worse.
The approval and availability of the treatment "pills" from Merck / Pfizer will provide the air cover to allow governments to resume "normalcy". So I'm hoping for summer'22 the VLSI symposium is back to in-person, but we'll see how CES and ISSCC play out before then. Vaccines are preventative, and the pills are for treating symptoms when you catch it eventually (and with Omicron, the likelihood that you catch it eventually is high). But that results in the same 1-2 punch for Influenza, where you can opt for an annual vaccine, and in the event you catch it, you take Tamiflu or equivalent. For Covid-19, you'll have a semi-annual vaccine schedule, same as visiting your Dentist, and if that proves insufficient, then you have the option of a treatment plan. Then the onus is back to being on the individual if they want to take these preventative measures or not. I can't see the era of daily testing continuing, or even testing at border entries, but hey - we're still asked to take off our shoes at the airport, so maybe this is one more joy of travel to look forward to.
 
I think the authorities like govt tasks forces, doctors, etc. knew about flurona two weeks ago and there was a fear. Now people seem much more relieved.
 
Keeping this on topic and summarizing:

1- Fab capacity for 200 mm flat-lined several years ago.
2- Not every IC can be cost-effectively made on 300 mm, either because of lot size or because customer qualification. Automotive is notorious for being slow when it comes to changes in manufacturing facilities, let alone wafer size AND geometry changes at same time.
3- When covid hit big time (March 2020), things were going swimmingly meaning 200 mm factories were near full. I am basing this on my company's sales. There was a retrenchment of 30% that lasted (in our case) until July 2020. This corresponds to car makers canceling orders, weakening economy and fabs reacting accordingly.
4- WOFO + bitcoin created all kinds of new demands that materialized in June-July 2020 time frame. Again, not everything is high-tech and that new Bose head set did not need 5 nm. Factories were near fully loaded right then, within the constraints whatever equipment they already had, + PPE shortages, labor etc.
5- You cannot make up time so here we are, everyone now switching to just in case so inventory building everywhere. This situation will last for another year, until enough new fabs are built and operational.

By the time car companies ate their seed corn, media started to report on it but in reality, it was a fragile system that was bound to fail, especially when China was re-born into everyone's bogeyman.

Incidentally, 95-98 MILLION cars are produced each year. Whatever Telsa has done is a rounding error to the overall situation.
 
So the question really is: When will consumer spending (demand) get more in line with manufacturing (supply)?

It already does. COVID unleashed a once in a decade spending splurge, similar to what happened in 2008, propelling forward the iphone.

Developing world is also finally maturing as a market. Xiaomi makes more money in India, than it does in China for example.
 
TSMC did not mention a chip shortage in their recent earnings call. They did say that customer inventories were increasing:

"Entering 2022, we expect the supply chain to maintain a higher level of inventory as compared to the historical seasonal level given the industry's continued need to ensure supply security. While the short-term imbalance may or may not persist, we continue to observe the structural increase in long-term semiconductor demand underpinned by the industry megatrend of 5G and HPC-related applications. We also observed the higher silicon content in many end devices, including automotive, PCs, servers, networking and smartphones. As a result, we expect our capacity to remain tight throughout 2022 as we believe our technology leadership will enable TSMC to capture the strong demand for our advanced and specialty technologies."

At some point in time the double and triple ordering will stop and customers will trust the supply chain again and start using up the excess inventory so a small wafer bubble may be coming.

TSMC also upped CAPEX to a record high $40-44B:

"Over the last three years, we have raised our CapEx spending from US$14.9 billion in 2019 to US$30 billion in 2021 as we invest in anticipation of the growth that will follow. During the same period, our revenue in U.S. dollar terms has increased from US$34.6 billion in 2019 to US$56.8 billion in 2021 or 1.6x and our EPS by 1.7x."

I would not say a glut is coming but I do believe TSMC is aiming for more market share when all of the post pandemic capacity comes online by putting pricing pressure on competitors. No one makes wafers more cost effectively than TSMC, not even Samsung, absolutely.
 
Back
Top