Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Semiconductor Shortage and Car Production

This site may earn commission on affiliate links.
... chip manufacturers have responded to increasing demand with investments in new capacity. Once those investments will be completed and supply will slowly, but surely, meet demand, analysts warn there is a risk of chip glut, ...
In May, South Korea announced plans to invest $450 billion to boost its chip production. The investment is channeled through several channels, including tax breaks, for companies that operate in the country.


The joint venture could attract investments of over $14 billion, though, for the time being, it’s not exactly clear how much Toyota and Mitsubishi are willing to pour into this project.

If this was a simple problem of lacking supply, how did it get so bad so quickly? How did the demand grow so much faster than the supply???
 
$2 billion won't pay to build a single facility,

The question I would ask is if Tesla could consider building a small Fab jst to produce these low end chips for in house vehicle production...

Given that they could set up a Fab around Austin and produce for tesla world wide planned to be 20 Million EVs per year by 2030.
The aim isn;t so much to save money but to de-risk supply, Tesla becoming partially self-sufficient opens up supply for others...

As these low end chips are really cheap, all Tesla would need to achieve is produce them for around the same cost as current suppliers sell them,

There is no risk of a bubble for Tesla, they know their projected production volumes..

However, this would take time to achieve even if they started today, and that can't do this every time these is a supply issue with car parts..

Still it would be handy to know timeframes and costs...
 
The question I would ask is if Tesla could consider building a small Fab jst to produce these low end chips for in house vehicle production...

Given that they could set up a Fab around Austin and produce for tesla world wide planned to be 20 Million EVs per year by 2030.
The aim isn;t so much to save money but to de-risk supply, Tesla becoming partially self-sufficient opens up supply for others...

As these low end chips are really cheap, all Tesla would need to achieve is produce them for around the same cost as current suppliers sell them,

There is no risk of a bubble for Tesla, they know their projected production volumes..

However, this would take time to achieve even if they started today, and that can't do this every time these is a supply issue with car parts..

Still it would be handy to know timeframes and costs...

Reminds me of when Musk wanted to build his own car carriers to solve the trucking shortage.

I think you missed the fact that the chips being discussed (which may or may not be the same as the chips actually short) are MCUs which are not so trivial to design and would involve a new instruction set and new tools requiring support, etc., etc., etc. There's a reason why people buy chips rather than make their own and the factory is only a part of it.

Just as with the car transportation shortage, perhaps time spent looking down the road would be a better investment than trying to factory build your way out of problems.
 
MCUs which are not so trivial to design
Unless they can license an existing design...

It is unlikely, but so was Tesla getting into battery cell manufacture, mining and insurance... or Tesla designing a FSD chip....

Merely being difficult, and requiring significant expertise and money, doesn't always stop Musk considering it...

But in this case it is a temporary problem with a non-strategic part of the business, and it is unlikely they could make a difference in time...

There is another reason why I mention this.... transitioning cars to a central wiring harness.. that is strategic,, and would save a lot of copper wire, cost and weight. My opinion is that needs chips with Ethernet connectivity and the ability to step down DC voltages as needed.

Hypothetically the central harness is 48V DC stepped down to 12V or 24v as needed... boards also need to support local wiring e.g. indicators, lights etc.. which are switched over the Ethernet.

Should this be required, Tesla may use standard chips on their own boards, i have worked for a company that did that, it is no big deal... But again this is an area where Tesla knows they will need a significant volume of chips in future.....

Cars are becoming more digital with every iteration... so chips are becoming important components, essential for car production....

I think they still may own some of their own car carriers.... I would not assume even that is off the agenda entirely, because it is intended the Tesla Semi has a Platooning mode where a single driver pilots a group of semis... Tesla using the Tesla Semi to transport cars would not be a major surprise.
 
Most of the chips for autos use older process hand me down fabs (except for chips like Tesla AI chip). Costs too much to build a new fab as car companies want cheap chips so have to use fabs that are already paid for. Also takes expertise. When I worked at Texas Inst, sold an old fab with the process manuals, procedures etc, and they couldn’t get it working without help from people who had actually run the fab.
 
Unless they can license an existing design...

It is unlikely, but so was Tesla getting into battery cell manufacture, mining and insurance... or Tesla designing a FSD chip....

None of those things are comparable as they are either a matter of inventing the things themselves or simply hiring the people they need to make it work or both. Licensing an MCU chip only solves part of the problem partly because you then have to license the many other chips you also need but also because you still need to build the fabs and hire the people. Just like with the car carriers, there was no one bottle neck to solve, it was many limitations that got in the way.


Merely being difficult, and requiring significant expertise and money, doesn't always stop Musk considering it...

He can consider anything he wants. Doing it is another matter. Look how long it took to make their own batteries, YEARS!


But in this case it is a temporary problem with a non-strategic part of the business, and it is unlikely they could make a difference in time...

Bingo! Now you are starting to understand. Unless this was an issue that was likely to continue or happen again, it's now worth the effort. Companies don't become experts at everything. The ones who try end up selling off the poor performers.


There is another reason why I mention this.... transitioning cars to a central wiring harness.. that is strategic,, and would save a lot of copper wire, cost and weight. My opinion is that needs chips with Ethernet connectivity and the ability to step down DC voltages as needed.

Hypothetically the central harness is 48V DC stepped down to 12V or 24v as needed... boards also need to support local wiring e.g. indicators, lights etc.. which are switched over the Ethernet.

48V autos have been mentioned for years. The conversion circuitry is not worth the bother since most electronics doesn't draw that much current. The wiring problem is more of having to run a wire for every sensor and button. Muxing those through a network with remote nodes is a good idea. No real need to drop 48V to 12V or 5V or 3.3V to power an LED.

Where 48V has an advantage is when everything in the car is built to run on 48V. That could work, but still doesn't solve much of a problem and may make the power off drain worse. Burning a kWh each day is pretty poor.


Should this be required, Tesla may use standard chips on their own boards, i have worked for a company that did that, it is no big deal... But again this is an area where Tesla knows they will need a significant volume of chips in future.....

Cars are becoming more digital with every iteration... so chips are becoming important components, essential for car production....

Becoming? It was just a couple of years ago the passives market saw a shortage because while the new production was focusing on mobile apps with ever smaller footprints (0402, 0201, etc), the growing auto market stayed with 0603 sizes since they give slightly fewer production headaches and the smaller sizes solved problems autos don't have, space. The result was production capacity didn't keep up with demand and it took a year for manufacturing to get new lines in place.


I think they still may own some of their own car carriers.... I would not assume even that is off the agenda entirely, because it is intended the Tesla Semi has a Platooning mode where a single driver pilots a group of semis... Tesla using the Tesla Semi to transport cars would not be a major surprise.

The "platoons" are not legal. The problem wasn't finding semi trucks, it was finding the licensed drivers and registered car carriers. The idea of building your own carriers to solve an end of quarter issue was absurd just as building a new fab to solve an MCU supply issue is absurd.
 
Most of the chips for autos use older process hand me down fabs (except for chips like Tesla AI chip). Costs too much to build a new fab as car companies want cheap chips so have to use fabs that are already paid for. Also takes expertise. When I worked at Texas Inst, sold an old fab with the process manuals, procedures etc, and they couldn’t get it working without help from people who had actually run the fab.

The fully depreciated fab cost is a red herring in this case. The two choices for Tesla is to buy an existing fab or to build one. In either case the fab won't be depreciated to Tesla.

Yes, it takes a lot of expertise... the guy doesn't get paid for swinging the hammer. He gets paid for knowing where to put the X to swing at to reference an old joke. It is mind boggling to consider they build a machine that can achieve repeatability to nanometers, tear it apart, ship it around the world, put it back together and get the same repeatability. Amazing! In centuries from now they will wonder how we did it like we can't figure out how they built the pyramids... except we will likely leave behind all the documentation. lol
 
He can consider anything he wants. Doing it is another matter. Look how long it took to make their own batteries, YEARS!
We can view that both ways, if something is important, Tesla don't mind spending years and a lot of money working on the problem..

In this case I agree MCUs are not important enough by themselves, to justify going to all this trouble,,,

The would need to be multiple reasons, both short term and long term...

In the case of the central wiring harness but understanding is the savings isn't just from a switch to 48V, it is a whole new way of wiring the car.. which should result in less wire being used even if the gauge remained the same.. They filed a patent for it years ago, so far nothing has happened... to me this implies it isn't a simple task...

For the DC-to-DC conversion it is a trade-off between conversion losses and a lower voltage... so I think they would often step down to existing 12v parts... the saving is the wiring comes from a local hub... the distance from the hub to the component can be small, the backbone has multiple hubs. Short wires also means lower resistance, perhaps offsetting conversion losses.

Maybe I am wrong and they patented the wiring harness with no serious intention to implement it. but for EVs saving weight is important, every bit counts..

EDIT:: My other thought is if the efficiency gains from lower overall resistance, in combination with the weight reductions may make this worthwhile. Perhaps Tesla can only get optimum efficiency with components they design and build themselves... But compared to the energy used in driving the motors, this is small beer.
 
Last edited:
We can view that both ways, if something is important, Tesla don't mind spending years and a lot of money working on the problem..

In this case I agree MCUs are not important enough by themselves, to justify going to all this trouble,,,

The would need to be multiple reasons, both short term and long term...

Just my point, building their own fab might be done for the high end chips some day, but even that would much more likely be done using contract fabs just like AMD and virtually everyone other than the contract fabs these days (oh, and Intel who isn't doing so well).


In the case of the central wiring harness but understanding is the savings isn't just from a switch to 48V, it is a whole new way of wiring the car.. which should result in less wire being used even if the gauge remained the same.. They filed a patent for it years ago, so far nothing has happened... to me this implies it isn't a simple task...

Distributed nodes in the wiring harness and using a 48V power bus are two totally unrelated concepts. Don't confuse them.

The distributed control nodes is not a new concept, the patent is because they want to do it in cars where the industry simply hadn't given it much thought yet. It is actually obvious to anyone in electronics providing other features in addition to reducing the wiring harness complexity.


For the DC-to-DC conversion it is a trade-off between conversion losses and a lower voltage... so I think they would often step down to existing 12v parts... the saving is the wiring comes from a local hub... the distance from the hub to the component can be small, the backbone has multiple hubs. Short wires also means lower resistance, perhaps offsetting conversion losses.

Again, two unrelated issues. Using a higher voltage has a single benefit, smaller diameter wire on the power rails. That's it. In ICE where it was originally considered, it would allow smaller gauge starter wiring and lower losses in that path (the reason you don't see 6V systems anymore). EVs don't have starter motors, so even less incentive, other than the fact that cars have so much more electronics now than in the 2000's when this was first brought up. It died because no one wanted to convert anything much less everything.


Maybe I am wrong and they patented the wiring harness with no serious intention to implement it. but for EVs saving weight is important, every bit counts..

How does it count any more than in ICE? Oh, you mean because of the range limitations. Ok, perhaps, but we are talking pretty small savings in weight. The distributed control nodes do a lot more than the 48V power rails in weight savings.


EDIT:: My other thought is if the efficiency gains from lower overall resistance, in combination with the weight reductions may make this worthwhile. Perhaps Tesla can only get optimum efficiency with components they design and build themselves... But compared to the energy used in driving the motors, this is small beer.

What "lower resistance" and what motors??? Oh, you mean some power savings which isn't actually there is significant in the face of the huge consumption by the propulsion motors. Valid point, especially since there is no lower resistance or power savings from either concept. The distributed control nodes will use more power if anything and I suppose you can make the 48V power wiring larger than needed so you don't save as much weight but save a bit of power in the I²R dissipation. You can do that at 12V if you want, too. The 48V system will require power to be burned at each voltage conversion, around 5% at the optimum efficiency point, potentially a lot more at other power levels. This is the sort of thing that raises the parked power consumption which is already too high.
 
Oh, you mean some power savings which isn't actually there is significant in the face of the huge consumption by the propulsion motors. Valid point, especially since there is no lower resistance or power savings from either concept.
Yes, that was my point... I can't see that the power savings are that significant..

Savings on weight and copper is significant, but cost savings on copper wire are offset by nodes...

Tesla will only do this if it is a good idea the solution would need to increase efficiency, hence lower power consumption, and result weight savings...

I suppose your points to raise the question of whether this is worth pursuing...

I also tend to agree they are very unlikely to build their own fab, so far we haven't been able to make a compelling case for that..
 
Different perspective on previous post. Toyota to cut North American production by about 180K vehicles in August and September.

Similar articles to the above two:
  1. Toyota slashes September output amid chip crunch, COVID resurgence
  2. Why Toyota's Production Cuts Are a Warning Sign for All Auto Stocks | The Motley Fool
 
Toyota sees further cuts to production:
Quote:
Toyota Motor Corp. willl slash global production again next month -- by 330,000 units -- as the pandemic and global shortage of automotive microchips continue to bite.

The total hit represents a 40 percent cutback from Toyota's original October production plan.

In announcing the reversal on Friday, Toyota said it will also take a bigger hit than expected in September. Toyota expects to lose another 70,000 units this month.

That adjustment comes on top of an August announcement, when Toyota warned it would lose 360,000 vehicles of output globally in September, including some 80,000 units in North America.

Unlike in last month's announcement, when Toyota kept its fiscal year global production target unchanged, Toyota said this time that it would lower its target to 9 million units for the fiscal year ending March 31, 2022. It had earlier planned to manufacture 9.3 million vehicles worldwide.
 
Not free Sept 19 article: A look at the roots of the chip shortage that has strained dealer inventory

Should a part the size of a rice grain and worth less than a penny slow down and even stop a 5,000-pound, $50,000, full-size truck from being sold? When that tiny part is a microchip that triggers an airbag, fires a fuel injector or commands a navigation system, the answer is yes.

In January, consulting firm AlixPartners estimated the global auto industry would lose nearly $61 billion in revenue this year because of the chip shortage. It has since amended that to $110 billion

Hopeful analyst predictions and soothing supply assurances from automaker CEOs and reps have proved to be optimistic. Even Toyota, which initially dodged North American production cuts, projected it would lose 60,000 to 90,000 vehicles in August and 80,000 this month. Honda also is warning its dealers of vehicle delivery cuts of as much as 40 percent in coming weeks.

... with launches of new vehicles blunted, used-vehicle prices took off. Today, used vehicles cost 15 to 25 percent more than a year earlier, if you can find them. Rental companies, which initially sold off their fleets thinking people would avoid shared transport because of virus concerns, are now scooping up even high-mileage models at auctions as travel restrictions ease and previously housebound folks go on vacations and visit relatives. Dealers at those same auctions are finding it difficult to buy low-mileage pre-owned vehicles.

Responding to the early chip scarcity, the Detroit 3 juggled several things. First, they prioritized chip supplies for more profitable, higher-volume models such as pickups and SUVs over slower-selling, low-margin vehicles such as sedans. Second, they reduced special options, such as navigation and advanced driver-assist systems, that required chips. Third, they built and stored vehicles without chips to be finished when supplies are reestablished. Some makers even considered shipping unfinished vehicles to dealers for retrofits. Still, even with these desperate moves, automakers such as General Motors, Ford and Stellantis were announcing plans to temporarily idle production lines.

Europe accounts for about 10 percent of global chip production, and the U.S. makes about 12 percent, down from 40 percent in 1990. Most of the remaining 78 percent is in Taiwan, China, Japan and South Korea. All producers have promised varying amounts of added production to cover global requirements.

Critically, the chips many automakers are looking for are not exactly cutting-edge silicon. Semiconductor companies are hesitant to invest much in increasing capacity in what are referred to as fabs and foundries for older, 200-millimeter silicon wafers, since in many cases this is 15-year-old technology.

Every week, industry expert estimates of lost vehicle production increase. AutoForecast Solutions keeps adding to its damage report. As of early this month, it was expecting the global industry to lose more than 8.5 million vehicles from production plans by the time the crisis ends. So it's fair to say analyst and expert crystal balls are fairly chip-free but full of dark clouds.
 
Most of those sitting on the sidelines are well aware that the market is being affected by a shortage of microchips, the research showed.

More would-be vehicle buyers are exiting the market, according to research posted Wednesday by Cox Automotive and Kelley Blue Book.

Most of those sitting on the sidelines, at least temporarily, are well aware that the market — which is short on inventory and high on pricing — is being affected by a shortage of microchips.

The August survey found that 48 percent of consumers were choosing to postpone a purchase, up from 37 percent in April.

Of those consumers who said they were stepping back from the market in August, 80 percent said they planned to be sidelined for between three and 12 months. In April, that figure was at 60 percent.

A shortage of microchips has constrained production of new vehicles, which in turn has led to increased demand and higher prices for used ones.

In the August survey, three-quarters of in-market shoppers said they would be willing to travel outside their local area to buy a vehicle. Most of those shoppers said they would be willing to drive 50 to 200 miles, but fewer than 20 percent said they'd go more than 200 miles. That question was not included in the April survey.

Also in the August research, 35 percent said they would shift from an imported brand to a domestic one, compared with 28 percent in the April survey; 32 percent said they'd switch brands in general, down slightly from 33 percent; and 31 percent said they would change which vehicle category they're shopping for, such as an SUV, crossover, minivan, compact car and so on. That compares with 19 percent in April.

Thirty-eight percent of shoppers in August said they would switch from a new vehicle to a used one, but only 18 percent said they'd go the opposite route. In April, 23 percent said they would go from new to used, while 16 percent said they would switch their search from used to new.

A little more than a third of shoppers in August said they would pay above the sticker price for new vehicles, compared with 42 percent in April.

More than half of shoppers in August were aware of the cause of the microchip shortage, compared with 46 percent in April. And nearly three-quarters said they were familiar with the effect it has had on the automotive market. In April, that figure was 53 percent.

"With a large portion of the in-market population now saying they plan to delay their purchase given the current market conditions, it will be interesting to see how that could impact the ongoing delicate balance of supply, demand and pricing across the industry," Vanessa Ton, senior industry intelligence manager for Kelley Blue Book, said in Wednesday's report.
 
Semiconductor companies tell car companies to start using newer chips.
Quotes:
“I’ll make them as many Intel 16 [nanometer] chips as they want,” Intel chief executive Pat Gelsinger told Fortune last week during his visit to an auto industry trade show in Germany.

Carmakers have bombarded him with requests to invest in brand-new production capacity for semiconductors featuring designs that, at best, were state of the art when the first Apple iPhone launched.

“It just makes no economic or strategic sense,” said Gelsinger, who came to the auto show to convince carmakers they need to let go of the distant past. “Rather than spending billions on new ‘old’ fabs, let’s spend millions to help migrate designs to modern ones.”

Most systems in cars are safety-critical and need to perform in practically every situation regardless of temperature, humidity, vibrations, and even minor road debris. With so much at stake, tried and true is better than new and improved.

“A lot of it just has to do with the fact that these are proven designs,” explained Gelsinger, who is currently campaigning for subsidies to build the most advanced chip fab in Europe.

“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.

If semiconductor suppliers like Intel and Qualcomm have their way, however, the days of the auto industry relying on these cheap commodity chips are numbered.