The global chip shortage is starting to have consequences

[1] https://www.bloomberg.com/news/articles/2021-04-29/apple-fin...

[2] https://unctad.org/news/shipping-during-covid-19-why-contain...

That really depends on how old your car is. I generally buy instead of lease, and run the car until it starts having regular problems. As a result, my car is 12 years old, and any new one from the last few years is a significant upgrade in terms of features. I don't really want a monthly car payment again, but I'm almost looking forward to when my current car starts having enough problems to make the upgrade worth it.

> Presumably, due in part to eviction protection, they have been able to divert all or a portion of their rent payments to keeping up with bad car loans for commute vehicles that often will not be needed.

There are gonna be far fewer evictions and/or mortgage defaults than people think, all those ppl who have been out of work are likely making more money right now than they were at the start of the pandemic just from unemployment, additionally bec they’re unemployed they may qualify for their states’ Medicaid benefits and food stamps, not to mention the free school lunches than many states have turned into a basically tons of raw produce and other various meal-making materials delivered/picked up each week. All those extra benefits mean the raw unemployment dollars go farther compared to a normal income creating an effectively higher $/hr wage than if you just look at the $300-600/wk(+state unemployment).

All this distills down to the fact that everyone has been flush with cash the entire time so much so that I know a few people in March 2020 who were behind on rent but due to all the aforementioned benefits were able to pay the rent they owed. Meaning that people having been making rent and/or mortgage for the most part and have probably been living beyond their (normal) means for the last year. If there’s gonna be any sort of correction it’s not coming until mid-2022 at the earliest.

Americans will pay their car loan before their rent.

In terms of bells and whistles? Yeah, not that much. But in terms of safety tech, there has been a huge progress in last 10 years and it’s not slowing down. New cars are not only getting harder to crash, but in case of crash they’re getting better and better at protecting you.

Safety is main reason why I update my cars every few years, even tho it’s pretty costly. But so is having my family seriously injured or killed.

Separately, I suspect that quite a few of the ppl who had rent relief had cars that, let’s just say, wouldn’t be highly sought in he used vehicle market.

Companies always piggy back on the current crisis to make changes or blame issues on that crisis that have nothing at all to do with the actual crisis...

come to think about it, governments also do this.

One vendor is having trouble finding any chip to design for. The software stack supports a couple of dozen STM32 chip series, but none of them are even remotely available. I've seen a lead time of over 11 months. Before COVID, pretty much every single chip was in stock almost all of the time.

Respinning a board isn't too difficult, but good luck doing that if there's no chip to respin it for.

If it's a big news story people are focused on it. And humans mistakenly attribute events to what the are focused on.

Trump was a great scapegoat for local, city, regional, state politicians.

All kinds of things can be blamed on climate change, systematic racism, etc. Basically anything that a leader can't fix but is in the news is a great scapegoat.

Not to say that these aren't real issue. They are, but once you hear how they are over attributed as causal you can't unsee it.

This is just not true. But when it happens, there is usually an announcement well ahead of time (e.g. 1 year), as well as a last time buy option.

Historically, there has never been a need to buy the chips of all the boards of a revision. If you are in the position of being able to buy all the components of all the boards you'll ever make for a certain revision, you're probably just above hobbyist level.

...

> if you aren't buying enough chips to build all the boards you want to build of a revision

Friendly advice:

If you are going to be a consultant to industry, don't post comments like this.

As someone who has been manufacturing tech products for over thirty years, my first reaction to your comment was "this guy doesn't have a clue". Then I looked at your site and was absolutely floored. My guess is you have lived in what I like to call the "SBIR distortion field", which is a domain that is very, very far from the realities of, say, a dog washing company. Not just because of usually just having to make one or a few of something (rather than 10,000), but also because of the financial dynamics of these programs --I have experience in that domain as well.

Your vision of how this dog washing-machine company should operate does not align with the realities of a business outside of the "SBIR distortion field". Companies don't have cash reserves to fill the warehouse to the brim with components and product, weather a storm, keep the business afloat and everyone employed simultaneously. On top of that, manufacturing at any non-trivial scale is such a cash intensive endeavor that cash must be managed very carefully. If you buy too much inventory you can end-up in financial dire straits.

The phase lag between spending money to manufacture a product and getting a return on that investment can be in the order of months, and that assumes a "linear" market. If you include R&D in that equation, it's even worse, years.

I experienced this personally back in 2008. I did precisely what you suggested above and filled the warehouse with some two million dollars in components and assemblies to get ready for sales of our new product. We had demand. In fact, the purchase of the components and assemblies was triggered by receiving a purchase order for five million dollars of this product. And that was just one customer. I didn't know better. I thought it was perfectly sensible to place large PO's for critical components that would cover us for at least a year and tool-up. We even bought a bunch of brand new CNC equipment to bring manufacturing of heat sinks and other mechanical components in-house in order to reduce our cost basis. In fact, interestingly enough given some of what you have on your site, I made the single largest purchase to date (at that time) from Osram's high power LED division. No company in the world had ordered that many high power LEDs from them.

And then the music stopped.

The economy came to a grinding halt.

Sales went to ZERO.

The five million dollar purchase order? They went insolvent when their bank cut-down and eventually cancelled their line of credit. Other orders from major companies were put on hold (we had a PO but were told they were not going to accept deliveries, so, don't ship). We went from having tens of millions of dollars in orders for that product and that year to, effectively, zero.

What was the end result? It was very rough. All of our cash was in the warehouse, on shelves, as components and assemblies we could not sell. We couldn't even get a loan to weather the storm. Nobody was buying anything, not at scale anyhow. We had to sell some of our component inventory for ten or twenty cents on the dollar just to bring in cash. It was worthless.

I had to take a second mortgage on my home and use credit cards to make payroll (big mistakes, both of them). We survived for two years on bread crumbs. And then I had to shut down the company. It too me years to even be able to talk about this episode of my life to anyone. It was horrible.

The two millions dollars I spent on "buying enough chips to build all the boards", as you put it (it was more than chips, but the example fits) was the single biggest mistake I have made in my business career. And this one cost me a business I built over a decade, starting in my garage with $5,000 to receiving a $30MM acquisition offer just as the economy took a shit (the offer was rescinded).

So, please, pretty please, with sugar on top, if you want to be a consultant, don't say anything unless you really understand it. In this case, you clearly do not. To someone like me --who has actually lived through many ups and downs in life and business-- such comments result in what I will call "less-than-favorable conclusions" about the author. This isn't good for a consultant, unless the consulting is in a domain that does not necessarily align with reality outside of something like the SBIR/academic domains.

It took years to recover, both mentally and financially. I eventually launched a new business, also in tech. Today we are facing having to manufacture 10K to 20K units per month of a new product. When we started design we picked readily available components and went on to design the product over about twelve months (real product design for scale manufacturing takes time).

Today, as we approach production requirements, we are being quoted anywhere from 40 to 50 weeks for some of the components. In other words, we can't even buy them. We are having to consider having multiple alternative designs to see if we can manufacture functionally equivalent versions of the hardware using different chip sets. This means all of our regulatory and safety testing --another thing you ignored-- (FCC, CE, TUV, UL, environmental, thermal, lifetime, etc.) has to be redone, not once, but likely four to six times (depending on how many versions we end-up with). It's a nightmare.

And, no, buying a million chips a year ago wasn't the solution. The cash drain would have resulted in people losing their jobs and possibly even going out of business again as sales levels last years went down some 80%.

You buy as close to just-in-time as you possibly can. This practice has gained acceptance over the years for a reason. Sadly, I happen to have learned the lesson the hard way. If you have to weather a storm it is far better to have cash in the bank than a warehouse full of worthless components that you can't turn into cash precisely because of the storm.

"A man holding a cat by the tail learns something he can learn in no other way". --Mark Twain

So true.

Companies hate holding on to inventory. Most companies now don’t hold on to any extra stock needed to manufacture their products.

This is great for your finances in normal times. But don’t complain when there is a shortage and you literally have zero slack for delays and your factory sits idle — that’s the well-known drawback.

You cannot support right to repair if you support the idea that chips go obsolete.

What you are suggesting is quite impractical, unless you are a really big business with cash, which can simply direct order components.

Small businesses, even in Shenzhen, a place inundated with supply chain abundance, always have to either keep running from one small wholesaler, to another searching for components, or pay n-times the price working with somebody like Arrow.

In my experience, you can't safeguard yourself against such things as a small company no matter what.

I worked in, and around OEM electronics since 2007, and things like having to redesign a product 4 times a year to accommodate a supply chain disruption were happening even in the best years. Nothing special with the current disruption besides the scale.

This is also the reason why Asian electronics makers have such short product lifetimes. It's usually easier just to sunset a product, and make an improved, and better version with newer components, than to fight against the always evolving supply chain. And I not talking about small companies, ASUS, Acer, MSI, and such all practice this.

I know few people running the Chuwi brand. What they do as a small maker is that the moment the get a good consignment of chipsets, and other parts, they spin a laptop model solely for that batch alone.

Then, they live off it until they get another good parts purchase, when they usually sell their component leftovers, or do few final runs if they can find people wanting to buy them for rebadging.

The entirety of small volume laptop industry spins around chipsets, and screens — hardest things to find for a small company.

I want you people to take a looks on a big difference in how companies in the West, and here handle the crisis: Western brands wail, cry, and wait for component supplies to resume, while Asian brands just keep releasing new products with parts coming into their hands, and making great cash from this shortage.

https://www.digikey.com/en/products/detail/xilinx-inc/XCVU13...

... as soon as my board hits production. If they give me any grief at DigiKey, I'll have my VP of Procurement fly up there in the company G5 to give them a piece of my mind.

He bought in two units to fix to have in stock since he already had a unit running his greenhouse.

I asked him why use Vic-20s when there were a number of controllers already on the market. He said the problem he never knew when the present unit may fail and the Vic-20 did the job of an $10,000 controller which he did not know how long it would take to get the present controller to work on his system and the present one works fine.

It has been decades since I did the work and I would not be surprised if the system is still working today because he was prepared.

We have already re-spun PCB designs to work in alternate ICs for some of the ones that we know will become unavailable; dual-footprint in case the originals come back. This took time and effort to design and verify and all that work could have gone into the design of new products, but it had to be done. This has put us some number of months behind our original planned dates for our new products.

The problem is that we can see shortages coming up which we cannot work around. We cannot buy enough parts, the lead-times are astonishing. At some point we are going to have to get lucky or halt production.

Turns out relying on a single region and in that region mainly on a single country for producing most essential goods wasn't such a good idea. Especially when said country is ruled by a rival regime.

In the EU the maximum authorized mass (US: GVWR) you can drive with a standard car license is 3500kg or 7700lbs. That means for some models of the F150 you do need a commercial license.

Example: Fiat Fiorino: https://www.hojeemdia.com.br/polopoly_fs/1.792949!/image/ima...

It used same Chassis as Fiat Uno https://quatrorodas.abril.com.br/wp-content/uploads/2019/08/...

So it was obviously a car, with the rear-part modified to carry cargo.

Meanwhile the 1980s F150: https://i.pinimg.com/564x/89/50/8a/89508a513b5076ef6413543f2...

It is obvious that thing is NOT a car, when you learned to drive in a UNO you can't expect a F150 to drive the same!

<rolls eyes>

You (statistical you as a HN user, not you personally) say this with one breath but one look at the B pillar of an 80s minivan and you'd start coming up with excuses for why it's not safe enough for your kid.

The same demographics that complain about cars growing do a hell of a lot more complaining about safety and efficiency. So the OEM adds space for crumple zones, thick strong doors, a bulbous front end that pedestrians can gently (lol) bounce off of. And the OEM makes smooth sculpted curves so the vehicle can punch the cleanest possible hole in the air. And then they realize there's no room for people, so they scale it up to make more space so we get a Civic that's bigger than a 90s Accord.

You can't have it all HN, so which is it?

What am I supposed to search for, small small trucks? Trucks that are actually small? Looking for cheap trucks still delivers large vehicles, just with fewer features: https://www.motortrend.com/news/cheapest-pickup-trucks-frill...

I even searched 'smallest trucks' and...guess what. The Honda Ridgeline seems like the most compact but they're all pretty chunky. https://www.web2carz.com/autos/buying-and-selling/8257/these...

If you suggest used trucks, that just shores up the point that manufacturers keep building bigger trucks. Please show us these small trucks of which you speak, even if they're more expensive or whatever.

/me waves from my tiny Corolla, slightly larger than my 2002 Prius, which a Camry has shrunk in size to meet relative to the 2009 model.

There are some situations where it’s great, but those features make the driver feel a false sense of security. Let the people feel like they’re entirely responsible so they’re more careful.

For example, widening the roads in the suburbs used to seem like a great idea - more space, less accidents, right? But that’s untrue. Narrower streets with trees blocking visibility on the sides are actually safer because the driver is forced to be more aware [1] [2].

Another good example is road markings and street/stop signs. Surely, having street signs and lane markings is safer right? Well, this is early on, but at least on city streets, it appears that’s not true either. [3] [4]

Now, to your point, it does appear that automatic car safety systems do make cars safer right now [5]. But those types of things are pretty new, so I wonder how long it’ll be until they have the same fate as those other safety innovations of the past. Where taking them away will make driving safer because people feel personally responsible, so they drive slower.

[1] https://www.thecalifornian.com/story/news/2015/06/14/studies...

[2] https://www.cnu.org/publicsquare/narrow-streets-are-safest

[3] https://gizmodo.com/this-street-has-no-lanes-signals-or-sign...

[4] https://www.nytimes.com/2005/01/22/world/europe/a-path-to-ro...

[5] https://www.forbes.com/advisor/car-insurance/vehicle-safety-...

Any data to back this up? Note that there is a difference between "always better" and "generally better". The former needs only one counterexample.

My car has on several occasions failed to detect that there is a car in front of me, and would have happily crashed into it. About once a year, it also suddenly applies the brake hard thinking there is a car in front of me when there isn't. That is quite dangerous - were there a car behind me it would have rear ended me.

One area where I strongly believe (yeah, personal feelies, no empirical study to back it up) misguided priorities have led to sub-optimal safety results is forward visibility. Some sibling comment mentioned "car didn't detect / erroneously 'detected' vehicle in front"... Since when do we need technology to "detect" vehicles in front of us; aren't we supposed to be able to SEE those, in order to be qualified to drive?!?

A: Since A-pillars have gotten as fat as your thigh, obscuring huge swaths of your field of vision.

Now sure, I understand that it's a good thing you won't get crushed flat if your car rolls over. But it's a bad thing if you crash into a vehicle, or run over a pedestrian, because you can't see out of your own vehicle. And how often do roll-over accidents really happen? My guess would be, a lot less often than collisions caused by impeded forward visibility.

(If my opinion that "visual aids shouldn't be necessary" in the forward direction makes me a "mental maths-advocating teacher", then so be it.)

And then half the cars can’t sync your phone without a dozen menu selections.

I think I need more description around "owning the world". Sure, those leisure focused companies are pocketing big dollars, good for them! But they don't really do much, they're only meaningful given the backdrop of the larger economy.

I don't see Morris Chang running to open an advertising shop, webhosting company, or mine bitcoins, thought the later was much speculated.

Coca Cola has a larger market cap than Pfizer. mRNA vaccines are definitely a lot harder to make than sugar water. At an individual level sales people are often paid more than the engineers in the org, despite the latter doing much more complex work and having specialized training.

The material handling systems are a sight to behold. Most may not be aware, but automation rates of 95% and beyond are feasible in these facilities. Many times, a lot can be moved through the entire manufacturing process without a single human touching or even looking at it. The amount of code and engineering around the material handling system is a mind boggling expanse of complexity. When million dollar product is flying around at 30+mph overhead on robots, you tend to take your time and do it right.

Add on top of this the same time domain concerns you have in global logistics, but with far more acute consequences. It's probably ~OK if a cargo ship is 15 minutes late to port. If lots miss a special process timeout by the same duration, you are potentially looking at millions of dollars in scrap. These intervals can be as short as 30 minutes. Imagine trying to schedule high priority lots through special tools, intermixing with unrelated (but also urgent) manufacturing nodes, while also allowing for preventative and break-fix maintenance on these same tools.

By far, the most complex parts of these semiconductor manufacturing operations are the business rules for the overall manufacturing environment, followed closely by the IT infrastructure required to tie together tens of thousands of hyper-complex technology systems. You better believe this is the one time it 100% makes sense to use pub/sub messaging.

Oh and don't forget about all the HF acid, ultrapure water, EUV lasers, hyper-scale cleanrooms, exotic power distribution systems, et. al.

I was trying to highlight, if you want to make a chip, you can, but it'll take a long time to build up the knowledge and skills. So maybe, you hire people with that knowledge and those skills. Of course, a lot of speed comes from not having to coordinate with lots of people. Small teams don't have as many communication channels, solo work has no communication channels.

There's probably a group of people and a set of supplies that could get a fab built surprisingly quickly. I don't think automakers are suffering from 5nm chips, they can get away with older processes. Good luck finding all those resources funding, coordinating, and building the process.

Perhaps not hard like discovering new process, but a _lot_ of work. It's not going to be a weekend project. This is what I meant by, It's going to be hard.

Almost like starting a "generics" business in pharma medication but for older chipsets.

I'm sure there's a great trade to be had in producing the lower end stuff.

Absolutely not. You just made this up.

Secondly, some legacy manufacturers of semiconductor parts lost money on their capacity-building investments during the dot-com burst. The semiconductor industry is brutal and there is a genuine fear that overcapacity will make it hard to deal with any bust that happens after this boom.

I mean, I agree it wouldn’t be good. But back to the 50s? Or even the “Iron Age”? That seems awfully pessimistic.

Like, I get that Intel is not doing great, but it’s only like a handful of years bad, not decades bad

TSMC is building a megafactory in Phoenix AZ which should help distribute that risk. It will take several years to complete however.

It's awesome how quickly the noise emerges from the deterministic rules.

Fuel (or really, oil) is a single commodity that everybody is after. If the price is low, and you're anticipating a shortage in the near future, you can buy now and sell later for a higher price. That's as simple as a market can get. buy low, sell high, whilst gambling on future price rises.

However, ICs are not a single commodity - you can't take whatever the foundry is putting out and slam it in place of whatever you actually needed. Each circuit is specific to its application. Sure, you could take a similar chip and rework your product to use the new chip instead of the old, but that isn't how a foundry works. They sell capacity.

Each company that wants their chips made will (or should) have done their forecasting for demand in the short-medium term. The problem is that those term limits have lapsed at a time when manufacturing is in a crunch. So now everybody needs new chips made, and nobody has the capacity to make them all.

Those that can rework their products to use similar chips that are already available should be doing so to maintain business. Those that can't do that, or are unwilling to do that, are paying through the nose to pay off other foundry customers to take their slots. meanwhile, there aren't enough chips to continue product manufacturing, so that's on hold until the new chips come out of the foundry and then everything can resume.

Prices will rise to cover the lull in production. And then businesses will see that people are willing to pay that amount, and so prices will not go back down.

If that isn't a more complicated financial market than oil, then I have no idea what is.

I'm not sure about n95s specifically but generally we saw mask prices hike, and then production increased quickly enough that there were masks for the majority of the population within a couple of months.

We can continue ad absurdum, but it's clear that more screens is added to our lives to:

1. Consume more content

2. As a result, see more ads

3. Finally, buy more stuff

I'm not against technology, but electronics producers have been going out of their way to continue growth. That's fine, but I'd love to have a robot doing dishes and cooking for me. Instead, there's a fridge that "Cameras recognize the food in your fridge so you can search for recipes based on what you have." [0]. That's such a marginal benefit.

[0] https://www.samsung.com/us/connected-appliances/#get-app

Doesn't really matter... if Fridge with a Screen makes 4x the profits or simply has a recurring revenue per unit where Fridge Without a Screen does not, then the market will become purely Fridge With a Screen.

First, sales numbers can only really ever tell you part of the story. If every fridge has a screen, sales numbers won't tell you about the demand for screenless fridges. If, at that point, some manufacturer tests a screenless fridge, the sales numbers might tell you about a lack of demand or might tell you about a failure in marketing. Additionally, if fridges with screens can be used to advertise to customers, get them to sign up for subscription services, steal their data, avoid the expense of bifurcating the line of fridges, or otherwise increase profit or decrease costs, they will be manufactured anyway.

Your experience with Lexus dealerships is not generalizable to the whole industry. The economic fallout of 2020 was not evenly distributed, and the only thing your personal experience tells me is that the audience for luxury cars may have had the better end of things last year.

[0] https://www.google.com/amp/s/www.forbes.com/wheels/news/2020...

Carvana and similar are such a better experience than dealing with a salesperson whose incentive and words you have to interrogate. I'd wager that salespeople will be needed only for high ticket price, high risk deals where trust is critical in getting the job done. Anything else that can be made transactional will be and the associated sales jobs will be lost.

If I was buying a new car the same would apply.

The only use case I can see is a notification when the cycle is done, but I think there are better ways to go about that than using an SoC.

I'm a programmer and honestly I can't wait for this IoT fad to die down a little. Sure, maybe it's cool to have LAN connected lightbulbs as a novelty product. But this kind of shit will never ever be the norm, simply because it makes no damn sense in terms of value proposition.

Any more understanding than that would be sub-optimal for shipping consumer products where cost optimisation is a primary concern as the salaries for more competent engineers would cost the company more.

You can see this effect in action with the explosion of “smart home” devices after commoditised internals were made available by the likes of Tuya. Suddenly your company only needed junior engineers who could skin the whitebox turn-key solutions and product designers who could design a moulded plastic enclosure around a standard set of postage stamp sized circuit boards.

What would you want to script?

wait what do you mean I still need to physically walk to the washing machine to load it this is bullshit

(Crickets.)

So don't hold your breath for this stuff to take off. Most people won't spend more money to buy something that will benefit someone else.

For now. The profits to be made on microtransactions and subscriptions from internet-requiring features are too gargantuan to pass up, to not become the new norm. And, of course, unblockable ads and tracking.

A fun recent example is a motorcycle emergency vest that stops inflating when you stop paying the subscription. An outlier for now, but the average tomorrow; The slope is real and it's coated with vaseline.

There are some problems for which banning isn't a solution, but there are some problems for which banning is actually a viable solution. Should we "learn" from the war on drugs and reverse the ban on CFCs? Should we avoid banning toxic substances from food?

You're gonna need a better argument than "activists who want to ban stuff should have learned from the war on drugs". You're gonna need to argue why banning cryptocurrencies is more like banning weed and not like banning CFCs.

The purpose is not to eliminate Cryptocurrencies, but to reduce the price of Bitcoin and other PoW coins such that mining (which is a misallocation of our civilization's energy and advanced manufactured goods) becomes significantly less rewarding.

Ethereum at least has a Governing body which is moving it to PoS. Bitcoin does not. Government regulation is the only answer.

edit: to head off the downvote brigade, I'm not advocating for this. Only pointing out that "banning" doesn't necessarily mean outlawing its use, but can also include making it unfavourable to use.