Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/the-five-biggest-mcu-suppliers-accounted-for-82-of-2021-sales.16228/
        )

    [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
)

The Five Biggest MCU Suppliers Accounted for 82% of 2021 Sales

Daniel Nenni

Admin
Staff member
New ranking shows some revenue gaps expanding between the largest microcontroller manufacturers.

Single-chip microcontrollers for embedded control and computing functions are ubiquitous and continue to be designed into more systems. Much of the new growth in MCUs is driven by embedded automation and the spread of sensors. The pervasiveness of MCUs was a key factor in suppliers being unable to keep up with the strong 2021 rebound from the 2020 global recession caused by the start of the Covid-19 virus pandemic.

After falling 7% in 2019 because of a weak global economy and then dropping 2% in 2020 due to the Covid-virus crisis, MCU sales rebounded with a 27% increase in 2021 to a record-high $20.2 billion. The 2021 surge was the highest percentage growth in MCUs since 2000. The average selling price (ASP) for MCUs climbed 12% in 2021—the highest annual increase since the mid-1990s. Production-constrained MCU shipments grew just 13% in 2021 to 31.2 billion units.

In the strong MCU recovery last year, the sales rankings of the five largest microcontroller suppliers remained unchanged from 2020, according to IC Insights’ new second-quarter update of its 2022 McClean Report service (Figure 1). The 2Q Update shows three of the 2021 top five MCU suppliers headquartered in Europe (NXP, STMicroelectronics, and Infineon), one in the U.S. (Microchip), and one in Japan (Renesas).

The Five Biggest MCU Suppliers Accounted for 82% of 2021 Sales.png


Figure 1

The five largest microcontroller suppliers develop and sell ARM-based MCUs. These companies accounted for 82.1% of worldwide MCU sales in 2021 compared to 72.2% in 2016—meaning the big keep getting bigger in microcontrollers. The increase of the biggest MCU suppliers has resulted from major acquisitions and mergers since 2016. The five biggest MCU suppliers are significantly larger than the rest of the top 10 in microcontrollers, according to the update report. For instance, the second half of the top 10 (Texas Instruments, Nuvoton, Rohm, Samsung, and Toshiba) accounted for $2.3 billion in MCU sales last year, or 11.4% of the market total. Outside the top 10, MCU suppliers had just 6.5% marketshare in 2021.

In 2021, top-ranked NXP in the Netherlands slightly widened its MCU revenue lead over second-place Microchip by $103 million. Microchip increased its sales lead over third-ranked Renesas by about $40 million last year, according new estimates in IC Insights’ 2Q Update report.

Fourth-place STMicroelectronics saw the strongest sales increase in the MCU ranking with revenues rising 35% in 2021, which nearly lifted the company past Renesas—putting it just $46 million behind its Japanese rival. Renesas had been the largest MCU supplier through the first half of the last decade but was passed in 2016 after NXP acquired U.S.-based Freescale at the end of 2015. Renesas’ marketshare in MCU sales stood at 17.0% in 2021 compared to 33.1% in 2011.

Germany’s Infineon remained in fifth place in the 2021 microcontroller ranking with sales that increased 22% to $2.4 billion—about $996 million less than ST in MCUs last year. Infineon moved into the top five MCU ranking after acquiring U.S.-based Cypress Semiconductor in April 2020 for $9.3 billion to expand further in automotive microcontrollers, power management, and other embedded systems applications.

Report Details: The 2022 McClean Report
The McClean Report—A Complete Analysis and Forecast of the Semiconductor Industry, is now available. A subscription to The McClean Report service includes the January Semiconductor Industry Flash Report, which provides clients with IC Insights’ initial overview and forecast of the semiconductor industry for this year through 2026. In addition, the second of four Quarterly Updates to the report was released in May, with additional Quarterly Updates to be released in August and November of this year. An individual user license to the 2022 edition of The McClean Report is available for $5,390 and a multi-user worldwide corporate license is available for $8,590. The Internet access password and the information accessible to download will be available through November 2022.

https://www.icinsights.com/services/mcclean-report/pricing-order-forms/

More Information Contact
For more information regarding this Research Bulletin, please contact Rob Lineback, Senior Market Research Analyst at IC Insights. Phone: +1-817-731-0424 email: rob@icinsights.com

PDF Version of This Bulletin
A PDF version of this Research Bulletin can be downloaded from our website at https://www.icinsights.com/news/bulletins/
 
Very likely noname Taiwanese/Chinese MCUs weren't even counted.
Obviously chips suppliers are selling more. But where did those chips go?

To highest bidders, which are often way more moneyed than car companies. I watched myself how certain STM32 models were auctioned at $100 per chip. That's from sub $1 prices.

Cars makers are some of the biggest MCU buyers, but there are many more buyers than them which were fine with double digit USD MCU prices.
 
Very likely noname Taiwanese/Chinese MCUs weren't even counted.


To highest bidders, which are often way more moneyed than car companies. I watched myself how certain STM32 models were auctioned at $100 per chip. That's from sub $1 prices.

Cars makers are some of the biggest MCU buyers, but there are many more buyers than them which were fine with double digit USD MCU prices.

Are car makers' profit margin razor thin and can't afford to pay more?

Does that mean the more semiconductors are adopted into cars, the more auto makers will be going out of business?
 
Are car makers' profit margin razor thin and can't afford to pay more?

Does that mean the more semiconductors are adopted into cars, the more auto makers will be going out of business?

They can buy MCUs at $20-$30 no problem for top range cars. It's $10000 econoboxes which only sell at $1000 profit. To maintain 10% profit, the price has to go up, by up to a half in some cases.

Clearly, practices like using one MCU just to blink a single button will go away.
 
They can buy MCUs at $20-$30 no problem for top range cars. It's $10000 econoboxes which only sell at $1000 profit. To maintain 10% profit, the price has to go up, by up to a half in some cases.

Clearly, practices like using one MCU just to blink a single button will go away.

Sooner or later some automakers who can't design and build car better will probably go out of business.
 
They can buy MCUs at $20-$30 no problem for top range cars. It's $10000 econoboxes which only sell at $1000 profit. To maintain 10% profit, the price has to go up, by up to a half in some cases.

Clearly, practices like using one MCU just to blink a single button will go away.
In the US at least, there's no such thing as a $10,000 new car. The Chevrolet Spark at $15,700 is as cheap as they go, and there aren't very many Sparks sold. In fact, in 1Q22 there were almost none sold.

The problem automakers have is the volume of chips they need. $100 for a $10 chip may not mean much when you need 1000 units. When you sell 750,000 Ford F150 pick-up trucks per year, and each one has 10s of chips, the "auctions" probably don't make any sense.

I agree that automakers will likely consolidate functionality into more capable chips and implement more features in software. But that will take a while to implement. Probably years, since I doubt they'll do it until new versions of vehicles come out.
 
In the US at least, there's no such thing as a $10,000 new car. The Chevrolet Spark at $15,700 is as cheap as they go, and there aren't very many Sparks sold. In fact, in 1Q22 there were almost none sold.

The problem automakers have is the volume of chips they need. $100 for a $10 chip may not mean much when you need 1000 units. When you sell 750,000 Ford F150 pick-up trucks per year, and each one has 10s of chips, the "auctions" probably don't make any sense.

I agree that automakers will likely consolidate functionality into more capable chips and implement more features in software. But that will take a while to implement. Probably years, since I doubt they'll do it until new versions of vehicles come out.

It's not even a matter of implementing things in software, but using 1 serial register. That's it.

The problem they are dealing with is called "modular architecture." OEM parts, each with own pack of MCUs, are bought off the shelf, and linked on buses.

If one blinky button comes from one OEM, and another blinky button from another, good luck integrating things.
 
It's not even a matter of implementing things in software, but using 1 serial register. That's it.

The problem they are dealing with is called "modular architecture." OEM parts, each with own pack of MCUs, are bought off the shelf, and linked on buses.

If one blinky button comes from one OEM, and another blinky button from another, good luck integrating things.
I think we're saying approximately the same thing, Paul, we're just thinking about two different solutions.
 
What I want to say is that they can design, and build better, and try using parts with consolidated IC functionality, but where would they get these parts? None of them can make parts in house. Nearly all parts production is outsourced.

I know people who work in car engineering for Toyota, a few Chinese who made it to Japan. Even in Toyota, custom parts are synonymous with anathema.

Just like in VLSI, 9 out of 10 companies will never do any complex IP, and only build cookie-cutter SoCs from off-the-shelf hard macros. And if they really have to, they will hire Indian/Bangladeshi outsourcers.

30 years ago, nearly all computer companies were able to make their computers from scratch. Now..., it is all coming in knock-down kits, and most brands have really became purely knock-down kit assembly business.
 
What I want to say is that they can design, and build better, and try using parts with consolidated IC functionality, but where would they get these parts? None of them can make parts in house. Nearly all parts production is outsourced.
I agree, and, with the exception of Tesla, I'm not aware of any automaker doing in-house IC development. I also agree that software development is usually out-sourced, and often to low-cost GEOs. The out-sourcing of critical software development even affects aircraft manufacturers. If my memory is correct, Boeing out-sourced the development of the notorious 737-MAX MCAS software. Oftentimes in mechanical engineering companies, like auto and aircraft manufacturers, complex and poorly understood technologies like IC and software development are out-sourced. Auto manufacturers, again excepting Tesla, even go to outside vendors for critical electronic systems like ADAS (e.g. Mobileye and Harman) and instrumentation / entertainment systems. I think auto manufacturers look to Boeing, Airbus, and Embraer for their models of how to build a large-scale manufacturing company for very complex products. Focus on the big picture, leave the detailed brush work to others. I know I can argue both pros and cons of this strategy.
I know people who work in car engineering for Toyota, a few Chinese who made it to Japan. Even in Toyota, custom parts are synonymous with anathema.

Just like in VLSI, 9 out of 10 companies will never do any complex IP, and only build cookie-cutter SoCs from off-the-shelf hard macros. And if they really have to, they will hire Indian/Bangladeshi outsourcers.
Agreed. But in their defense, the libraries of hard macros available have become very impressive. I haven't heard much about quality concerns either, which frankly surprises me.
30 years ago, nearly all computer companies were able to make their computers from scratch. Now..., it is all coming in knock-down kits, and most brands have really became purely knock-down kit assembly business.
I used to work for one such from-scratch vendor, a BUNCH company, who even did their own microprocessors, chipsets, and I/O chips. Overall, I think the trend to IC development consolidation is a good thing. Those old proprietary chips were not often impressive, even in chronological context. The low-volume economics were difficult to straddle even in the 1980s. As recent as 15 years ago some vendors, like HP, were still doing their own microprocessors (PA-RISC), and HPE still designs networking chips for their Cray Slingshot HPC interconnect. IBM still designs both Power and Telum microprocessors, among other chips for their Z16-series mainframes. Dell and Lenovo, who I think are on your mind with your kit assembly comment, are not the whole industry. Yet.
 
What I want to say is that they can design, and build better, and try using parts with consolidated IC functionality, but where would they get these parts? None of them can make parts in house. Nearly all parts production is outsourced.

I know people who work in car engineering for Toyota, a few Chinese who made it to Japan. Even in Toyota, custom parts are synonymous with anathema.

Just like in VLSI, 9 out of 10 companies will never do any complex IP, and only build cookie-cutter SoCs from off-the-shelf hard macros. And if they really have to, they will hire Indian/Bangladeshi outsourcers.

30 years ago, nearly all computer companies were able to make their computers from scratch. Now..., it is all coming in knock-down kits, and most brands have really became purely knock-down kit assembly business.


40 years ago automaker Ford tried to design and build chips inhouse. But it failed.

"In 1982, Ford Motor company created a new subsidiary (Ford Microelectronics Inc.) to oversee research, development, design and manufacturing of their own Semiconductors for use in their production vehicles, and for aerospace applications for their other subsidiary, Ford Aerospace & Communications Corp. (producing satellites / defense products)

Ford Microelectronics was based in Colorado Springs, CO. They initially worked with silicon wafers for design and manufacturing of their custom integrated circuits. But in 1985, they would build a $33M factory to attempt to be one of the first companies to manufacture gallium arsenide semiconductor circuits in high-volume. Competing with Ford Microelectronics for the commercial gallium arsenide semiconductor market were Microwave Semiconductor (Siemens), Tachonics (Grumman), GigaBit Logic, and Vitesse Semiconductor."

Source: https://www.chipsetc.com/ford.html
 
Last edited:
They can buy MCUs at $20-$30 no problem for top range cars. It's $10000 econoboxes which only sell at $1000 profit. To maintain 10% profit, the price has to go up, by up to a half in some cases.

Clearly, practices like using one MCU just to blink a single button will go away.
Not with a $20 - $30 MCU, but there are plenty of 50-cent MCUs. The problem now is availability due to the lack of capacity on trailing nodes.
 
This has been discussed to death, but, what is the deal with 8 inch fabs and $1 automotive parts? Why doesn't NXP, to name one 8 inch fab operator, sign a contract with a Foundry, get some 300mm trailing edge capacity, on 65nm say, and fill the unmet demand in the market? Is this going to happen, but just hasn't happened yet because of frictions (mask production, validation, contract expirations, something like this?), or 65nm Foundry not being cheap enough, or 65nm Foundry reliability or ???
 
Not with a $20 - $30 MCU, but there are plenty of 50-cent MCUs. The problem now is availability due to the lack of capacity on trailing nodes.

Say, NXP, and STM clients move to Padauk next month, and the inventory story will repeat. Once they will raise price, they will keep it high knowing they are working for a car company.

This has been discussed to death, but, what is the deal with 8 inch fabs and $1 automotive parts? Why doesn't NXP, to name one 8 inch fab operator, sign a contract with a Foundry, get some 300mm trailing edge capacity, on 65nm say, and fill the unmet demand in the market? Is this going to happen, but just hasn't happened yet because of frictions (mask production, validation, contract expirations, something like this?), or 65nm Foundry not being cheap enough, or 65nm Foundry reliability or ???

They can't redesign parts. A part designed for a certain STM32 made on 200mm, will have to use the exact same die until the end of the EOL which was pre-agreed, and put into contract long ago.

So, a part maker may have a blinky button with STM32 on 300mm already, but old cars will have to use old parts without any change. In large part due to car model needing a new registration even if you replace a single nut in it in some jurisdictions.

Does it makes sense? Hell no, but Western countries are littered with such self-destructive policies destroying the industry. No wonder people call it a "legal minefield."

It greatly favours outsourcing just because anybody incompliant will be shut down in the West for good, but a Chinese OEM will just change its name, reopen in the same place a day after, and restart shipping to Western clients in one month time.

The self-destruction of the industry through legislation is a glaringly obvious problem on a national security level of significance. More than any single chip fab. TSMC's biggest fear in Arizona is not even the water running out, but whether "this, and that will sue us to death"
 
Last edited:
Why doesn't NXP, to name one 8 inch fab operator, sign a contract with a Foundry, get some 300mm trailing edge capacity, on 65nm say, and fill the unmet demand in the market?
You're assuming that there's some extra trailing edge capacity at 65nm, which may be a big if.

TSMC's earnings call in January 2021 shed a little bit of light on this: (I've been looking to see if they've commented on it more recently and can't find anything yet.)

Robert Duncan Cobban Sanders - Deutsche Bank AG, Research Division - Director
Yes, I just got one question, actually. Just could you please then comment more on the wafer shortage situation and how severe it is at present? Which node are you seeing the shortage most acute? Is it 65-, 90-nanometer, 0.11, 0.13, whatever it is? And how far out are you essentially booked out at some of these nodes? And do you think there's wafer upside to what you're pricing at these nodes?

Jeff Su - Taiwan Semiconductor Manufacturing Company Limited - Director of Investor Relations
Okay. So Robert, your question is on the tightness or shortage in the wafer. He is asking, is it at particular node such as 65-nanometer, 90-nanometer, 0.13, how short it is and how long it will last.

C. C. Wei - Taiwan Semiconductor Manufacturing Company Limited - CEO
Robert, most of the shortage actually is in the mature node. It's not in the 3-, not in the 5- or 7-nanometer per se, but in all the mature node, especially in 0.13 micron, in 40-nanometer, in 55-nanometer, in those area.
 
So the shortage is due to: Inflexibility of long-term supply arrangements; and at TSMC there there is a shortage of capa from 130 to 50nm.

This raises another point: Foundry capa is shared capa, as this blog said:

”Car companies now have to compete with every other sector of the electronics industry for wafer capacity in overseas foundries.”

 
So the shortage is due to: Inflexibility of long-term supply arrangements; and at TSMC there there is a shortage of capa from 130 to 50nm.

This raises another point: Foundry capa is shared capa, as this blog said:

”Car companies now have to compete with every other sector of the electronics industry for wafer capacity in overseas foundries.”


One side of the problem I found out recently, is that the number of SKUs with automotive certification is very low.

Their newer MCUs may be on 300mm, but they will keep the price as these guys can't simply jump to non-certified models out of a sudden.

So these MCU makers will just get a massive margin boost.
 
I've been looking to see if they've commented on it more recently and can't find anything yet.
Twitter press coverage of the European session of TSMC Tech Symposium quotes as 65/55 is the most over-subscribed node currently:

However, their capacity build-out only seems to be for 28/22nm and newer (based on their comment of 50% increase for 28nm capacity, at least). So customers will have to consider migrating at some point. Serious question on what is driving the uptick in demand though on these specific nodes, because one could assume that many of these "10yr+" auto grade supply parts have been in production already for some time, and demand forecast shouldn't have gone 2-3x just for this year. Was there a shift in ECUs or similar controller specific to EV management that would correlate (maybe a specific customer's battery charging monitor IC)? I would have though those were on older nodes still. It can't all be Apple's VR demand, or CoWoS.
 
Back
Top