Hacker News Reader

Hacker News

Comment by 11thEarlOfMar | Hacker News Reader

Full thread

11thEarlOfMar

Several factors:

1. There was real and relatively sudden increase in demand:

a. COVID-driven in part, all students and many additional employees had to buy computers to study/work from home,

b. a., driven by video, lead to increased demand on Internet & telecom infrastructure.

c. Electric vehicles consume an order of magnitude more chips than ICE cars. EV demand is also skyrocketing.

2. There were fabs (semiconductor factories) shut down due to COVID. This one in Malaysia [0] makes power chips and was particularly problematic for ICE auto makers.

3. The first indication of problems manifested in chip lead times pushing out. As buyers started to see more and more of this, they started increasing orders and soon adopted a hoarding mentality, similar to why many of us couldn't find toilet paper in stores early in the pandemic.

Result: As a personal anecdote, a chip that sells for $0.47 from that fab in Malaysia recently sold for $75 on the broker market.

Note that no, the state of the art fabs cannot be re-outfitted to make legacy chips. The old equipment they were developed on is generally no longer available and the profit from making those inexpensive chips would not justify the capital spend on new equipment.

[0] https://www.reuters.com/article/infineon-results-malaysia/in...

View on HN

Regarding #3: We had an FPGA that is normally in stock or at most a few weeks lead time push out to a 56 weeks lead time. (Side note: This is for machine used to in chip manufacturing... uhm, oops.)

What happened then is that the vendor prioritized larger more expensive parts. They use more silicon per part but the over all profit per wafer is higher. It's not a terrible decision, it's possible to move designs to an over provisioned FPGA but usually not the other way. The lead time for higher end parts started coming down, not to normal, but less than a year. Meanwhile lead-times for the low end parts kept getting longer.

This scenario is very reminiscent of the leadup to the Innovator's Dilemma. The dinosaurs grow taller and fatter seeking and thriving on the juicy foliage higher up the tree, while the scrappy little mammals claw away at its soft underbelly.

Nobody's going to fabricate semiconductors in their garage. But couldn't someone make a modest profit without first investing $2.0e10, like Intel? Or is it just that the profitability scale is such that if you had an extra $2.0e7 to invest, you're better off buying Intel stock?

I don't think that you can build a new semiconductor factory for $20 million (2e7), especially not when there is a chip shortage happening and you need factory equipment containing such chips.

State of the art fabs retail for US$10 Billion. TSMC, Samsung, Intel & China can afford to build them.

(for context, Freedom Tower cost US$2 Billion.)

I think in 2022 it’s more like 15B. 10 sounds very early-2020.

Perhaps you could do it at a 1000nm node, at very low volumes.

EBeam lithography will get you sub 10nm, but it's one at a time.

This is close to how they make the mask-sets, very slow.

There is the volume vs cost problem. You could put a fab together for under a million - but to do so your would trade labor for automation. A simple chip that a large fab can make for a profit (amortized) selling for 5 cents each would cost you more than $50,000 each to make. If you only need one chip that price is worth it, but most users of a chip are thinking a lot more and it typically turns out that getting a large fab to do your part is not only cheaper, but faster. Still you can do the highly manual process of making a chip if that is what you want to pay for.

It's a little weird to me. I see in the comments 50 cent chips going for 75 dollars. The old process chips _seem_ diy able - http://sam.zeloof.xyz/first-ic/ Obviously it would take me, like, a year or two to get anything to work. But I'd guess there are folks out there that could spin up far far faster.

I guess I don't understand why there isn't much, or any, labor intensive fabs combining on line, with a plan to automate away parts of the line as money comes in.

I suppose, the chemicals are so toxic, and the skills are so rare you can't get any production for months, even at the small scale. and perhaps even at the crazy high prices it's still not worth taking on the risk.

I know I don't know what I'm talking about. But golly it does _seem_ like I could make chips in a couple of years, and there's bound to be thousands of people out there that could do it in weeks. It seems like the kind of thing you could scale out with more labor and training, and improve process with automation as you go.

edit

Most trying to understand. Is it too toxic? are the skills so rare? Is the risk that high? Is that what's keeping low end chips from using a more expensive process even though they command higher prices - it's not enough to offset?

The $0.50 for $75 is a short term situation that will resolve once the fab's pipelines get filled again. We buy >1,000,000 per year and had no trouble getting them until the fab shut down. We'll be fine soon (looks like they've opened production again).

The primary costs of starting up a fab are 1. facilities, 2. equipment. Facility per square foot is likely the most expensive you'll find due to the appurtenant systems required for air flow and purity, a variety of gas handling systems, ridiculous electrical power systems, possibly gas plant systems, water purification and then treatment systems for both air and fluids exiting the plant. Some truly scary toxins are used and produced and have to be contained or captured and neutralized.

In terms of equipment, you'll need at least one of each of 40 types. Lowest cost is about $500,000. Highest is likely $25,000,000 (at the geometry you're talking about). Average in the area of $3,000,000 - $5,000,000. You'll need a cadre of engineers and scientists to keep it all running, and 2-3 years from breaking ground to producing product. If you did buy one of each tool type, you'd have most of them sitting around doing nothing most of the time, since the process times vary dramatically. In reality, you're looking at $500,000,000 minimum to build a fab that can produce $0.47 chips. Seriously, you don't want to do that.

Thank you. That makes a lot of sense.

I had some misconceptions around how versitile auto manufacturers are. I'd sort of assumed they'd be very good at turning a low volume, low quality process into a high volume, high quality process. I'd also sort of assumed they'd have a ton of resources on hand, like space and power, the and the ability to build a lot of the required infrastructure.

I suppose they are far far more specialized and reallocating resources like that is the kind of thing you'd have to do right away, make a big commitment, that's basically pointless.

They are experts in iron, and processes, but they still would have a lot to learn, and it is unlikely they could ever do better than TMSC who isn't stupid about manufacturing. Maybe they could find something, but so can the others.

The commitment just to play is very large, it probably isn't worth it. Even if you had proof of this would happen in 2015 and so enough time to build I doubt any would build s fab (though they would buy chips to store)

Yup. makes sense.

Wasn't thinking 7nm cutting edge, more along the lines of op amps or maybe a 70's era microcontroller - I had the impression lots of chips are used, and some are real fancy, but some are just old and, since they didn't order them, aren't being made.

I'd thought if someone was willing to roll back the clock 30 or 40 years - that tech might be cheaper and more realistic to bring in house. But yeah, there are already plenty of people that do that well, so it's really just an expensive boondoggle.

Definitely not something you can just put together in a few months.

Sure old process can be done,but the labor is high enough that you need to sell for more than 75. For simple stuff you still need well over 1000 each to make it worth it.. that is simple stuff. I know a few years back the government was paying around 60000 for Pentium level complexity chips (this was for a classified program so I doubt I was given full details)

Something is very weird with intel. Compared to fang pe around 30 or nvidia, amd around 100, intel has 10. Which combined with long term stable decently high profits is insane for an am tech company. Intel has been offered gigantic subsidies to build american fabs, and more recently just fabs period, but they arent, and at least one of these is approaching the point where they might get sued for not having even tried.

The answers I hear range from a deeply embarrassing and complete loss of competency on all levels, but primarily on the fab side, to an utterly incompetent and out of touch leadership. I favor the latter explanation as intel has been underperforming for nearly two decades now. My best guess is that intel has entered its cash cow phase, and will simply phase out product after product as they become obsolete, ending up as a half dead zombie barely supporting a few rare products in a few decades. Problem is, even if thats what is happening, their stock is surprisingly cheap. So it seems analysts seem to know something important I am completely missing.

Not to say they are the only company that fucked up, amd has been offered the same subsidies for building more fabs, and we arent talking 2020, we are talking for more than a decade, and they also havent for very unclear reasons. Amd also fucked up and completely missed machinelearning due to being unwilling to make a convenient interface or even just halfassed support for cuda. But they did that while nvidia went, ah fuck it, well let amd get every major console, and basically only make gpus with graphics ports as a side project.

Amd and nvidia is havent acted optimally and both have failed to exploit what seems like low hanging fruit, but they are highly valued in a way which matches market demand (which has been increasing quickly since 2014, not just 2020), and their improving technology. Intel however, does not even appear to have tried. Performance improvements have been incremental at best, and some "features" like the built in gpus are so terrible they mostly degrade performance.

No argument on the fab side, but that's easy to outsource. Samsung, TSMC, and Global Foundries will happily make Intel chips.

Alder lake is a good example, from what I can tell it's a market leading core, generally faster than the AMD competition, and a good basis to make future products. Future Intel GPUs seem likely to compete on price/perf, at least when any GPU from the last 3 generations seems to sell out instantly.

> Samsung, TSMC, and Global Foundries will happily make Intel chips.

And how long will it take to "port" those designs to another fab?

It's not a matter of "make install."

Yes, everything comes together with tools, but every one of those tools and its inputs is verging on full-custom and those designs were tweaked with the characteristics of those tools in mind.

The folks who know how to do that work are currently working on new products (either for Intel or someone else), so are you willing to slip those products?

Quite a bit faster than building a fab for a leading process and making a new product on that fab.

Alder lake seems plenty competitive (even market leading) for today, AMD has nothing big (like zen4) due anytime soon. Moving the next gen to some other fab in time to replace alder lake looks quite feasible.

Yes, Intel can do new designs for other fabs.

The question is when it makes sense to port old designs to other fabs.

FWIW, I'm not convinced that it takes less time to bring up a new design on a new fab than it takes to bring up an old design on a new fab. (The exception being a new fab that is identical to the old fab.)

> I favor the latter explanation as intel has been underperforming for nearly two decades now.

Remember "Only the paranoid survive"?

Well, people steeped in Intel took that to heart. So, when they retired or died, a lot of them had no heir apparent and their knowledge went with them.

1) technically, someone is fabricating semiconductors in their garage, but as regards the world economy your point is still valid: https://hackaday.com/2021/06/29/garage-semiconductor-fab-get...

2) the equipment used in older fabs, is in many cases not even made any more. It's much like why the vinyl record manufacturing took a decade or more to ramp back up when the demand for vinyl records began going up instead of down. Nobody was making that equipment any more. The older fabs don't use the same equipment as the new fabs, except less of it; they use quite different equipment, and in some cases newer equipment actually wouldn't be able to make the older chips. I don't work in semiconductors, but I did from 1989 to 2004, and worked on several cases of trying to move old processes into newer fabs. It's not easy or quick, and in some cases you have to move the old fabs' equipment into the new fab.

Also, the Texas storms caused hard power cuts to multiple chip factories, erasing, in the best case, a full month of chip production as they slowly restart the factories [1]. These chips are, btw, generally built on such old tech that Intel is refusing to fab them [2]. So I would have thought a chip was a chip, but apparently car chips can only be sourced from a handful of specific fabs.

Car manufacturers are clearly going to start building huge inventories of them though:

> “Because of a 50-cent chip, we are unable to build a car that sells for $50,000,” said Murat Aksel, head of procurement for Volkswagen Group, during a press briefing in Munich last week. [2]

Ouch. Though honestly, this probably ought to result in some execs being fired... If you sell 100k cars a quarter, imagine telling the board that you saved the interest on $200k and the rent on a small room and that's why you're not selling a vehicle.

[1] https://www.reuters.com/business/autos-transportation/texan-...

[2] https://fortune.com/2021/09/17/chip-makers-carmakers-time-ge...

> Ouch. Though honestly, this probably ought to result in some execs being fired

It should, but it won't because everybody failed the same way.

It was just like the 2008 banking crisis. Lots of people saw it coming, but reasoned "If I predict doom and mistime it, I'm an outlier and get fired. If I simply keep my head down and cash out and it all goes pear-shaped, I'm simply one of the masses and nobody will notice me."

Can confirm. Anyone working in supply chain is now told to buy 1-4 year supply of as many of the critical chips needed as they can. It used to be all about how JIT could it be. When the whole chip supply chain shifts from JIT to “get all you can while you can” it causes the shortages. 2nd hand market is seeing astounding mark-ups.

Does that mean most companies will be using 2020 generation chips until 2024?

Sounds they'll dump it in trash eventually and it'll be a terrible waste of resources

Companies in my industry have products they sell for 10-20 years. The chips will get used. Product updates often keep the same chip, and the same chip is used across multiple product lines.

EOL for these chips usually means a bunch of firmware engineers in sustaining need to scratch their head and read 10-15 year old documentation to find a viable replacement.

For the vast majority of things I buy that use chips (obviously not consoles, phones and computers) they could easily be using chips from 2010 and the only difference might be a slightly higher electricity draw.

Where I worked that I participated in the board design process, we'd get "lifetime buys" of chips on a many year cycle anyway for design stability reasons. Who cares if the microcontroller is using a couple more watts than it could be from aggressively respinning if it's sitting next to a 500w motor?

Chips are performing an astounding number of tasks these days. Microwaves, vacuum cleaners, toasters, etc. None of these applications require advanced chips and could settle for something from 20 years ago.

If everyone buys 4 years of supplies, that exacerbates the problem. The supply chain bullwhip effect will be massive.

For the next 3 years, fabs will be swimming in profit. After that, I expect a major correction.

Will the fabs have the discipline to actually use the extra cash injection wisely?

They're the best positioned to do that, they're one of the last classic long tail, high capital investment portions of the industry. Each node is only competitive for a short while, but they have multiple nodes in the development pipeline because of how long it takes, and a misstep on a single node can devastate the whole company if you're not in the #1 spot originally.

So the answer is we don't know it, depends on the management currently at the fab, but they have the right external incentive structure to handle it well.

Everyone is suggested to place scheduled orders for the future. Nobody (that I’ve heard) is suggesting that you immediately purchase and carry four years of inventory. People in this market are generally very intelligent, and there have been a number of booms and busts in semiconductors in the last 30 years.

> c. Electric vehicles consume an order of magnitude more chips than ICE cars. EV demand is also skyrocketing

Source? My impression is just the opposite. Batteries and an electric motor is a relatively simple device compared to a gas motor. So you replace a fuel pump (and related sensors), cat converter (and sensors), transmission (and sensors), and a hugely complex (1000s of parts) engine (and sensors). Managing maximum efficiency, torque, and HP across the entire RPM range while minimizing emissions is quite an intensive process. Not to mention the interaction between the engine and the (compute controlled) transmission. Especially these days when the simple rules are gone and managing turbos, heat management, etc means that you might give maximum HP, but only for 5-10 seconds or even idling unneeded cyclinders, stopping the engine at redlights, etc.

Tesla in particular as moves many distributed chips into a highly integrated single board, various people (like Monroe on Youtube) claim the Tesla model3 has a market leading level of integration replacing dozens of distributed CPUs with a custom board/ASIC.

I used to work in EVs (and tangentially still do). They really only require some power electronics - gate drivers and IGBTs - in addition to the normal bunch of chips all over the vehicle. So even if demand was "skyrocketing" rather than being a tiny fraction of the market, they shouldn't impact the demand for chips on regular cars.

EVs require 2.3x the chips of ICE. (not orders of magnitude, apologies for the hyperbole)

https://www.idtechex.com/en/research-article/ev-power-electr...

Interesting, and weird. Although that's measured by $, not per chip, per per mm^2 of silicon. I suspect electric cars (being newer designs) tend to use much newer chips, where the traditional ICE manufacturers are often using many generation old processes and paying rock bottom prices. So during shortages the cheap chips don't get made as fabs migrate production to the most profitable/newer chips.

Based on the tear downs I've seen I believe the 2000+ chips in the model S/X, but would be surprised if a model 3/y even had as many as 1000 (the number in a ICE vehicle). The model S/X were crazy complex, CPUs for each door lock, each window motor, each windshield motor, etc.

As a result VW, Ford, GM, Chevy, etc have to halt production lines, while Tesla has a record year.

> The model S/X were crazy complex, CPUs for each door lock, each window motor, each windshield motor, etc.

A lot of ICE cars have similar complexity in these systems. It's much easier to put a module in each component and just send messages on the bus to control them. Most newer cars also detect objects in the path of the window and will stop/reverse the window to prevent pinching.

While EVs might have more $ in chips, I also suspect they have a lower chip count as more indicators and interactions are handled on screen. Meanwhile, most ICE vehicles still have analog gauges, analog gear shifters, physical gear selection illuminators, etc. and all of these need chips.

I compared a 2019 Model 3 to a BMW i3, Chevy Bolt, and a Nissan Leaf (as well as a fair number of ICE cars like subarus, bmw 335, VW Golf R). I did notice that the entertainment system/nav/music, etc where worlds apart. The Tesla felt like a current smartphone, pinch/zoom, fast UI, easy to use, and pretty good speech recognition. The others felt like 5-10 year old phones, similar in feeling to even my ancient nokia 770 with Maemo mapper.

Monroe (on youtube) did a breakdown and said that the only other place he say CPU/ASIC/Sensor integration like on the model 3 was in satellites. Apparently there's a large CPU with a ton of pins that handles quite a bit of cars needs. Even the usually Bosche supplied ABS/Stability management/traction control functionality has been merged into the Tesla chip.

> Even the usually Bosche supplied ABS/Stability management/traction control functionality has been merged into the Tesla chip.

That's interesting... that's usually it's own ECU on most cars.

I can see why they might do it this way. ABS/SC in ICE cars tend to focus mostly on the braking system. Since most ICE cars are two-wheel drive with an open differential, there's not much that can be done from a stability point of view besides cutting power. Even AWD cars with computer-controlled differentials have response times that are too long to really be of much use in an emergency situation.

EVs, OTOH, can apply or reduce torque to the wheels much faster than even the brakes can respond. And being able to not only apply stopping force to a wheel, but also apply torque to another wheel makes for a totally different approach to stability management.

Here's an interesting anecdote: Tesla cars have traction control that monitors the differential in rotation speed among the four tires. If you put new tires on the rear and leave worn tires on the front, the rotation differential between the inside rear and outside front on a tight curve is enough to fool the traction control into thinking the worn front tire is slipping and will pulse the brakes on that wheel only. It's pretty unnerving.

Put new tires on the front and the problem is solved.

This happens on my BMW too (2011 3 series). I’ve also had traction/stability control activate when hitting a large bump due to the slight change in wheel speed as each wheel goes over it.

2.3x is a base 2 order of magnitude, so you were technically correct.

I suspect it's more the case that electric vehicles ship with more "smart" features (touchscreen consoles, mobile phone connectivity, etc.) which in turn leads to larger chip requirements.

Dunno, I don't think so.

Streaming music, maps for nav, top down views that integrate numerous camera/utlrasound etc sensors, onstar (and similar), traffic updates for maps, bluetooth connections/app for phone, etc seem common on pretty much all new cars. At least they were in the subaru, honda, bmw, and VW cars I was test driving. A center console mounted screen for entertainment/maps/music is becoming very common on pretty much all new cars I looked at.

Also 4: There's been underinvestment in the sector for years now. Lots of money got plowed into stuff like social media, crypto/blockchain & ML but semiconductors just weren't sexy to the investors.

I don’t really buy this idea that chip fabs were being starved of money because of blockchain. Do you have any evidence to back your claim up?

It's not so much that chip fabs were starved by blockchain as much as the fact that many investors generally chase high short-term low-risk yields, which primarily results in major investments being in software-based ventures. Hardware can yield quite large gains too, but the upfront risk is much higher and. the growth rate is slow (if at all nonzero) for a long time before actually growing.

The chip fabs of Intel, Samsung, TSMC, and a couple other companies were not being starved, no. But, nobody was starting new semiconductor fab companies like they did in the early days of Silicon Valley (hence the "silicon" part of the name). That hasn't been happening for a couple decades now.

> The chip fabs of Intel... were not being starved,

Maybe, but consider that Intel stock was pretty flat for a long time in the 2000s/2010s especially compared to the amount of money that was flowing into the FAANG stocks and other social media companies during that time. If Intel's stock had rising at something like those rates they probably would have been able to invest quite a bit more into fabs.

Companies don't invest off their stock price, they invest off their balance sheet, or by borrowing money. All these companies have plenty of cash at hand and can borrow at historically low rates.

Venture capital investment is flashy and gets the press, but there's lots of other money that wants more predicable returns. There was around $600B of VC invested in 2021, but Intel alone has $23B cash on hand, and US pension funds alone (which traditionally invest in safe sectors) has over $35T in management.

This isn't true.

Most semiconductor companies are public companies, with deep pockets and plenty of cash on hand. They invest based on forecast demand.

> Note that no, the state of the art fabs cannot be re-outfitted to make legacy chips. The old equipment they were developed on is generally no longer available and the profit from making those inexpensive chips would not justify the capital spend on new equipment.

Interesting. I had assumed that the difference between state of the art fabs and older fabs was that state of the art fabs can make smaller things, and perhaps more geometrically complicated things, but that they don't have to. If they wanted to make a design that only uses bigger things and simpler geometry they could.

Where does this assumption break down?

State of the art machines are used to make state of the art high-value chips. FPGAs, GPUs, CPUs, RAM, SRAM

As those chips run through their life-cycle, their value drops. After 15 years, the price the chips can sell for will drop 90% or more.

If there is a spike in demand at that point, there is no ROI on purchasing brand new equipment.

The demand spike is seen as temporary. Hence, the chip makers are telling the buyers to design out the old chips and upgrade to new. Many companies can do this. But some are constrained by industry regulation (aerospace) or industry practice (semiconductors).

I've mentioned before the irony that companies that make semiconductor equipment are struggling to ship more tools because they can't get enough (of certain chips) to complete building them.

While this is mostly correct, it is worth noting that FPGAs are mostly 28nm and 20nm processes, and have just started shipping on 16nm[1]. That doesn't require state of the art. The FPGA market is comparatively small though so doesn't affect the market dynamics much.

[1] https://www.xilinx.com/about/generation-ahead-16nm.html

>Interesting. I had assumed that the difference between state of the art fabs and older fabs was that state of the art fabs can make smaller things, and perhaps more geometrically complicated things, but that they don't have to. If they wanted to make a design that only uses bigger things and simpler geometry they could.

>Where does this assumption break down?

Why would they want to? There is heavy and overbooked demand for the 'smaller and complicated things' that make more money, and add the extra cost of the design to make the 'bigger and smaller' things, and it's a no brainer.

They wouldn't want to. What I'm curious about is if they can without retooling or purchasing equipment. Say due to chip shortages affecting the US military the government invoked the Defense Production Act to require some company with only state of the art fabs to make an older chip.

From what I've read, the way a layer on a chip is made is a coating is deposited on the surface that is sensitive to light at certain wavelengths. They then use light of wavelengths that the coating is sensitive to in order to transfer a pattern from a photomask to the surface, making some change to the coating at those places where the light is not blocked by the photomask. Then they can treat the surface to remove the coating at those places were it was changed by the light.

They then can deposit thing onto or remove things from those parts of the surface not still protected by the coating. Then they can use something that can remove the coating. The result is a layer on the chip with things deposited or removed right where they wanted them to, as determined by the pattern on the photomask.

Repeat for more layers.

The major improvements from generation to generation are the ability to use shorter and shorter wavelengths of light for the photolithography which allows for transferring patterns that have smaller features, and maybe the ability to make thinner layers so you can make chips with more layers.

I don't see anything in the photolithography process that requires using small feature size. If for some reason you wanted to build an older chip that only used much larger features, I'd expect that it would just be a matter of using the photomasks for that older chip, and setting the layer thickness to what the old design called for.

I think the deal breaker is "without retooling".

For one thing there's a lot of steps involved in photomask -> repeat. If your masks are all built for a smaller wafer size, or a different focal length setup or physical process diameter, then they literally won't fit in the die, or won't focus correctly.

You'll run into similar issues all up and down the chain. If your doping and prep all assumes a smaller wafer or different geometry, you're out of luck either physically because things don't fit, or because your chemistry needs to be redone. Your various deposition, bake and cure steps will necessarily also assume a process geometry, because some of the processes diffuse out from the center, or at different angles relative to some source that's bombarding them.

It's sort of akin to saying "well if we can build a vw bug they can build a model-T". Generations of equipment changes make things incompatible.

As someone with no experience in this area, I think you're ignoring the option of making the equivalent parts on the newer equipment.

The design costs would be higher (based on what I read, the new methods are more complex and costly), but if you're legally required to do so ... this would possibly be faster and cheaper then redesigning the factory. Larger wafers, larger yields, etc ... once working they could probably meet the required output quickly(month long delay?), provided the demand / legal punishments are great enough to justify the costs.

Only the most critical layers use the most advanced litho tools. The other layers use n-1 and n-2 litho tools and print larger geometries. You are correct that it is not a problem to print larger geometries.

However, there is more to it than just litho. Many thin films, implants, etching, polishing. A product is designed to a specific fabrication process flow. The real showstopper is that the process flow guarantees the electrical parameters of the transistors and passive devices will be with a certain range -- the product circuits are designed based on these parameters' values. A different process flow has all different electrical characteristics and the product won't work without redesign.

CMOS devices scale in all 3 dimensions, including gate oxide.

It is quite unlike, say, PCB fabrication, which is way more flexible with regards to layer stack-up.

Good list. Also, Taiwan drought and Texas freezing impacted some fabs last year.

And the fire in Ajinomoto's factory that makes Ajinomoto Buildup Film, used in high-end chips, briefly cut production of those.

I vaguely recall a shortage of epoxy resins as well. (For packaging chips, and for printed-circuit boards). It seems like more went wrong than usually does in 2020-21, even leaving aside the pandemic.

Non EVs also need a lot of silicon for infotainment, driver assist, etc. it’s not just EVs demand that is skyrocketing. All cars are.

True, but I believe there is also a missing factor: When people wasn't able to get outta home, demand for cars plummeted. And whoever made ICs as for cars just had the opportunity to fill the gap from automakers with more lucrative parts.

Ironically enough demand for cars went up during COVID. I can't tell you why, but I know it's a fact. Lack of supply is certainly a factor but I think there's probably more going on. Go check how insane the market is right now. It's nearly impossible to buy a new car, and used cars are selling at way above the notional blue book value.

Demand did go up, after the market adjusted to projections that with everyone in lock-down, demand for cars would go down. Demand for in home entertainment was projected to go up. Auto makers sold their capacity to people who were projected to use it in lock-down friendly ways. Once reality set in it was to late to get the capacity back.

We sold two cars and bought one new one. Remote work and a move to a rural area meant we had more cars than we needed, and that we needed to replace our city cars with something more rugged/utilitarian.

One thing I heard from a guy working on chip for a big German car company was that they used to have a decades old contract with a major Taiwanese chip manufacturer which had extremely good terms for the Germans. In particular with terms to delivery times, and fees for late deliveries, and price. This deal had apparently been used as a framework for the purchase of not just the then used chips, but pretty much all latter ones too. In retrospect the deal seems to have been at the very least partially politically motivated by Taiwan, but the chip manufacturer could not cancel the framework agreement without significant penalties. There was however an option for the car manufacturer to cancel quite freely, and when the news of the pandemic hit, they did. It seems like no one in the department quite understood just how extremely good the conditions had become over the decades, and after they canceled it, and decided they wanted the same deal back... Well lets just say they are at the back of the line despite paying substantially more than they used too.

I became 3400 dollars richer buy selling few old unused stm32 reels

Reply on news.ycombinator.com