Why are current gen motherboards so expensive?

M

mda

2[H]4U
Joined
Mar 23, 2011
Messages
2,207
Ok, I know costs everywhere are going crazy, but why are current gen motherboards so expensive?

I picked up my X570 Strix E at the local, USD equivalent of $250 (not in the US), and the X670 Strix E is more than double the price now. I read somewhere way before the 7000 series launch that the pair of X670 chipsets that come on those boards are cheaper than the single X570 chipset, so that shouldn't be the problem either.

This isn't only the case for X670, but also B650, Z690, Z790 and even the lower end B660 boards.

Implementing PCIE5 or multiple NVMEs surely isn't cheap, but pricing on boards with PCIE4 and not so many NVMEs like some B650s are pretty high too. Not sure having 5 NVMEs over X570s with 2 or 3 NVMEs would drive up the price by so much either.

Is there some secret sauce in these boards or are the VRMs 3x as expensive that will cause the total board price of the same tier to go up by 2x? I could have bought 2 X570 E (upper midrange boards) for the price of the current X670E TUF board, which sits a lot lower on the product satck.

Not trying to start anything geopolitical apart from what is already implied in my first sentence / what we already know, but this discussion is started as I am genuinely curious as to why the costs have gone up by a whole lot more than the current levels of inflation.

DDR5 costs are coming down and Intel 12/13th Gen CPUs are priced very aggressively for the performance they give, and I've been reading/watching rumors on 7000 series getting a price cut. But all the motherboards are still priced crazy expensive, and I'm not really so sure why.

Thanks!
 
Last edited:
#1 and most important, most obvious reason: People will pay. Maybe you won't. but someone will. These are new parts compatible with ONLY other new parts. they are a luxury item at the moment.

#2 PCI-E 5 and DDR5 require VERY strict timing and trace design, requiring better signal processing and isolation which often means additional layers of the PCB, additional SMDs like diodes and capacitors to make sure the signaling is clean and durable.

but mostly #1.
 
More power, more options, higher frequencies.... it takes more to build them. And of course, that imaginary word, inflation.
 
I wonder if in not your US country if there was not some somewhat recent exchange rate that could affect newer imported item more than older ones ? All speculative has well, because on US Amazon a Strix is currently $500 USD (depending by how much more than a double the $250 we are talking about), while the 570 strix E seem to $330 and B650-E $350, if you are in say the UK it could be massive.
 
Last edited:
cjcox said:
More power, more options, higher frequencies.... it takes more to build them. And of course, that imaginary word, inflation.
Click to expand...
All those things have been true for past generations as well, so that doesn't really explain the price hike.
 
Phazer Tech said:
All those things have been true for past generations as well, so that doesn't really explain the price hike.
Click to expand...
Was the change ever been that fast, PCI gen 3.0 was still common on x470 and intel 10xxxx system and now we are on gen 5.0 with 4 time the speed

1: 2003
2: 2007
3: 2010
4: 2017
5: 2019
6: 2022
7: planned 2025

Not many things double 2 time fast like that from 2017-2019, that connectivity (maybe more pushed by superbe density-need on the datacenter side) that get down to the deskptop is maybe getting a bit overkill for the moment for regular PC users (let alone if we multiply by 4 against quick, has a plus maybe almost everything will use only 1 or 2 lanes, turning a regular CPU in a potential workstation)

Not that dissimilar for Ram, DDR-2 standard topped at 8,533 MB/s, DDR-3 at 17,066, DDR-4 25,600, DDR5 already at 57,600 and they had support for them quite fast (at least by now on expensive intel side), not able to quickly check but would not be totally surprised if that did not followed on the power distribution side of things.

And obviously inflation was not like that since most people have a sense of motherboard pricing.
 
My 300 dollar z690 edge Wifi ddr5 doesn't even have a clear CMOS button. My 5 year old x99a sli plus does. Go figure. Let me go find my CMOS jumper how lame lol
 
KazeoHin said:
#1 and most important, most obvious reason: People will pay. Maybe you won't. but someone will. These are new parts compatible with ONLY other new parts. they are a luxury item at the moment.

#2 PCI-E 5 and DDR5 require VERY strict timing and trace design, requiring better signal processing and isolation which often means additional layers of the PCB, additional SMDs like diodes and capacitors to make sure the signaling is clean and durable.

but mostly #1.
Click to expand...

This....

This falls under one of the core concepts of Economy 101, Supply & Demand. "Demand" is the "willingness to buy" that the market dictates.
If not enough sheep buy, then the market is thrown out of wack.. and the costs have to come down. If the other direction, this shows the product is more desirable and costs will go up.
 
Because, like nvidia, they have realized people will pay more regardless. You watch, when inflation goes back down, lots of companies will still be trying to charge the same prices for stuff when it's ~2% vs. 8%.
 
Furious_Styles said:
Because, like nvidia, they have realized people will pay more regardless. You watch, when inflation goes back down, lots of companies will still be trying to charge the same prices for stuff when it's ~2% vs. 8%.
Click to expand...
Inflation isn't like "scalping". Once prices go up, salaries are (ideally) increased, valuation of currency goes down, but generally speaking, unless it's a temporal spike, prices don't decrease or at least do not decrease by much. Just the way it is.
 
Furious_Styles said:
Because, like nvidia, they have realized people will pay more regardless. You watch, when inflation goes back down, lots of companies will still be trying to charge the same prices for stuff when it's ~2% vs. 8%.
Click to expand...
Why company does not charge more, 2% more if inflation is 2% ? How can inflation be more than %0 without prices of stuff going up ?

I feel you mean if inflation goes negative 8% companies will still be trying to charge 2022 price, but by definition they would be failing at it.
 
LukeTbk said:
Why company does not charge more, 2% more if inflation is 2% ? How can inflation be more than %0 without prices of stuff going up ?

I feel you mean if inflation goes negative 8% companies will still be trying to charge 2022 price, but by definition they would be failing at it.
Click to expand...
Meaning if inflation is -6% for 2023, companies will still try to charge the same prices they did in 2022 when it was over 8%. Then they just pocket the difference.
 
cjcox said:
Inflation isn't like "scalping". Once prices go up, salaries are (ideally) increased, valuation of currency goes down, but generally speaking, unless it's a temporal spike, prices don't decrease or at least do not decrease by much. Just the way it is.
Click to expand...
Have you not been to the grocery store recently? Read any headlines in any newspaper? This is very likely a temporal spike.
 
Furious_Styles said:
Have you not been to the grocery store recently? Read any headlines in any newspaper? This is very likely a temporal spike.
Click to expand...

Inflation due to adding currency to the currency pool is necessarily permanent.
 
Furious_Styles said:
Meaning if inflation is -6% for 2023, companies will still try to charge the same prices they did in 2022 when it was over 8%. Then they just pocket the difference.
Click to expand...
So the 2% was a typo ? If the inflation is -6% in 2023 that would mean by definition they would not be able to charge 2022 price, they would be trying like always to charge has much as possible but failing, has a whole price would be down by 6%.
 
Axman said:
Inflation due to adding currency to the currency pool is necessarily permanent.
Click to expand...
Tax and other action can remove currency from the currency pool and if no added currency is added and good&service production rise you could see some deflationary action, no ?
 
Furious_Styles said:
Okay, but how much percentage-wise of the current inflation spike is due to that?
Click to expand...

I'm not sure, but I think the Fed doubled the number of dollars in circulation in the past two years. In a closed economy, that would double the prices of everything (and double wages, eventually) but it's not a closed system and a lot of other countries use the USD as a reserve currency.

LukeTbk said:
Tax and other action can remove currency from the currency pool
Click to expand...

You have to close banks and destroy currency to cause deflation, or increase other supplies. Like when AMD was flooding the market with 580s and they were $100-$150.

But it would have to be every industry tied to the USD, so...

I am sticking with my original position. Today's motherboards have the most wires. Seriously, they're way more complicated than motherboards from even five years back. They have more layers, more traces, tighter power requirements, way, way more bandwidth.

Even a budget board today is a high-end workstation board from a few years ago.
 
Axman said:
You have to close banks and destroy currency to cause deflation,
Click to expand...
Or just rise some lever to make lending rarer, like the big recent rise in interest rate.
Axman said:
I'm not sure, but I think the Fed doubled the number of dollars in circulation in the past two years. In a closed economy, that would double the prices of everything (and double wages, eventually) but it's not a closed system and a lot of other countries use the USD as a reserve currency.
Click to expand...
I think circulation would require a big "" here, reserve for example not being really what we would think has in circulation, I am not sure how much clear data we have, in reserve note, US note and currency:
https://www.federalreserve.gov/paymentsystems/coin_currcircvolume.htm
18% jump over 2 years, similar to the 17% inflation since 2019

Those M0-M1 supply type:
https://fred.stlouisfed.org/series/BOGMBASE
started to go down since fall 2021

But do not ask why exactly, apparently it is more complicated than looking at those to have an idea of circulating money:
https://fredblog.stlouisfed.org/2021/01/whats-behind-the-recent-surge-in-the-m1-money-supply/

It would be a lot about how things move around on balance sheet more than just amount of money on the streets.

has for price never going down to previous you are almost absolutely rate, deflation is rarely accepted and would be hard to see much motivation for it. I would imagine then just being happy going back to a 2% inflation rate and why not ? What would be the point to try to get back to previous pricing and forcing wage down to achieve that...
 
Furious_Styles said:
Have you not been to the grocery store recently? Read any headlines in any newspaper? This is very likely a temporal spike.
Click to expand...
Grocery prices were going up a lot pre-pandemic (I'm the grocery shopper in our household). Time will tell of course, but have a feeling you could be wrong. This "temporary" spike has been going on for quite some time, which is why I believe this to be in the same vein as in the past behavior wise (talking inflation). But again, time will tell. Right now, it just keeps going up as well, it's not plateauing, not yet.
 
It comes down to a lot of things.

1.) Inflation. We've seen massive inflation and it's impacted every industry. Everything is more expensive than it was two years ago. In some cases, by a significant amount.
2.) PCIe 4.0 and 5.0 have much more stringent requirements than PCIe 3.0 did. With PCIe 4.0, this required either additional PCB layers or the use of a different PCB material which is more expensive than the more common ones. Most companies actually chose the latter rather than the former.
3.) Given the nature of the PCIe lane configuration and HISO on the Intel side, there is a great deal of PCIe switching that goes on. This is fairly expensive to do. Anytime you have insufficient lanes to run everything on the board all at once, you have to rely on PCIe switching at some level in order to reroute lanes from say SATA ports 5 and 6 to the PCIe x4 slot at the bottom of the board, or to run a third NVMe device.
4.) As memory frequencies increase, trace layouts for memory slots becomes more important. DDR5 slots are essentially dual channel per slot. Running two DDR5 DIMMs is like running four DDR4 modules. You have a much more complicated trace layout for this.
5.) CPU VRM designs are advancing and the costs for these keeps going up. In fact, the VRMs contain some of the most expensive components on the board. On ultra-high end motherboards, you may have more than double the VRM's that a low end board would have and they are rated for considerably higher power output. You could be talking about comparing 4-8x 55A phases to 12 or more 90A power stages. Older designs may also use phase doublers which increase the cost dramatically as well. There is no free lunch with this. Even low end boards need more powerful VRM's than they did a few years ago for today's multi-core CPU's.
6.) Hardware monitoring and fan control requires specific dedicated IC's that are designed for that purpose. Today's boards have more of this stuff than ever before.

And this is without even getting into higher end boards and what they have to offer. They'll have some fluff features you don't need of course, but they also have nicer heat sinks and aesthetic touches that aren't free. They'll also have additional IC's for monitoring, more complex and more expensive voltage controllers, external clock generators, additional fan headers, thermal sensors, water flow sensors (more IC's), additional USB controllers or built in USB hubs, dual BIOS ROMs, dedicated IC's for BIOS flashback and the list goes on. In some cases you get fancy OLED screens and other things.

The motherboard practically is the computer. Everything is integrated into it these days. Add your own CPU, RAM, storage and video.
 
Dan_D said:
It comes down to a lot of things.

1.) Inflation. We've seen massive inflation and it's impacted every industry. Everything is more expensive than it was two years ago. In some cases, by a significant amount.
2.) PCIe 4.0 and 5.0 have much more stringent requirements than PCIe 3.0 did. With PCIe 4.0, this required either additional PCB layers or the use of a different PCB material which is more expensive than the more common ones. Most companies actually chose the latter rather than the former.
3.) Given the nature of the PCIe lane configuration and HISO on the Intel side, there is a great deal of PCIe switching that goes on. This is fairly expensive to do. Anytime you have insufficient lanes to run everything on the board all at once, you have to rely on PCIe switching at some level in order to reroute lanes from say SATA ports 5 and 6 to the PCIe x4 slot at the bottom of the board, or to run a third NVMe device.
4.) As memory frequencies increase, trace layouts for memory slots becomes more important. DDR5 slots are essentially dual channel per slot. Running two DDR5 DIMMs is like running four DDR4 modules. You have a much more complicated trace layout for this.
5.) CPU VRM designs are advancing and the costs for these keeps going up. In fact, the VRMs contain some of the most expensive components on the board. On ultra-high end motherboards, you may have more than double the VRM's that a low end board would have and they are rated for considerably higher power output. You could be talking about comparing 4-8x 55A phases to 12 or more 90A power stages. Older designs may also use phase doublers which increase the cost dramatically as well. There is no free lunch with this. Even low end boards need more powerful VRM's than they did a few years ago for today's multi-core CPU's.
6.) Hardware monitoring and fan control requires specific dedicated IC's that are designed for that purpose. Today's boards have more of this stuff than ever before.

And this is without even getting into higher end boards and what they have to offer. They'll have some fluff features you don't need of course, but they also have nicer heat sinks and aesthetic touches that aren't free. They'll also have additional IC's for monitoring, more complex and more expensive voltage controllers, external clock generators, additional fan headers, thermal sensors, water flow sensors (more IC's), additional USB controllers or built in USB hubs, dual BIOS ROMs, dedicated IC's for BIOS flashback and the list goes on. In some cases you get fancy OLED screens and other things.

The motherboard practically is the computer. Everything is integrated into it these days. Add your own CPU, RAM, storage and video.
Click to expand...

Then there are people like Vlogger MLID who are claiming motherboard manufacturers on the AM5 platform have marked up X670e boards an extra 20% for the quarter just because they can. This is despite the platform having two major cost-saving features: 1) the switch to using a dual chip chipset; and 2) reserving the installation of on-board paired power AND reset buttons for only the highest tier and penultimate tier motherboards. Look how many ~$500 boards only have a power button or have neither, nevermind the $350 and cheaper X non-e boards and B series boards.
 
Teenyman45 said:
Then there are people like Vlogger MLID who are claiming motherboard manufacturers on the AM5 platform have marked up X670e boards an extra 20% for the quarter just because they can. This is despite the platform having two major cost-saving features: 1) the switch to using a dual chip chipset; and 2) reserving the installation of on-board paired power AND reset buttons for only the highest tier and penultimate tier motherboards. Look how many ~$500 boards only have a power button or have neither, nevermind the $350 and cheaper X non-e boards and B series boards.
Click to expand...
The chipset being cheaper is a factor, but it doesn't necessarily mean the motherboard will be cheaper. X570 was expensive as it was literally a copy of the I/O die from the CPU with some added stuff. It was built on an expensive process as well. However, a two chip solution will potentially require additional trace paths and those are routed to a different part of the motherboard rather than all going to one place in the case of a singular PCH. I'm not saying some of the cost increase isn't artificial. I'm sure that manufacturers are using current conditions arising from inflation as an excuse to increase costs more than they need to. Even so, some of this is due to the inflation we've seen over the last two or three years.
 
Dan_D said:
The chipset being cheaper is a factor, but it doesn't necessarily mean the motherboard will be cheaper. X570 was expensive as it was literally a copy of the I/O die from the CPU with some added stuff. It was built on an expensive process as well. However, a two chip solution will potentially require additional trace paths and those are routed to a different part of the motherboard rather than all going to one place in the case of a singular PCH. I'm not saying some of the cost increase isn't artificial. I'm sure that manufacturers are using current conditions arising from inflation as an excuse to increase costs more than they need to. Even so, some of this is due to the inflation we've seen over the last two or three years.
Click to expand...
But can you explain the deal with the power and reset buttons?! I'm sick of building pcs on motherboards that lack them.
 
Teenyman45 said:
But can you explain the deal with the power and reset buttons?! I'm sick of building pcs on motherboards that lack them.
Click to expand...

You'd be in the tiny percentage of people that want or use them. The only use they have is if you have an open test bench, or are testing the board outside the case before installing it, and then never using it again.

If you need them that badly, do what the rest of us do and wire up a short header with a momentary switch on it, or two if you want both power and reset.
 
GiGaBiTe said:
You'd be in the tiny percentage of people that want or use them. The only use they have is if you have an open test bench, or are testing the board outside the case before installing it, and then never using it again.

If you need them that badly, do what the rest of us do and wire up a short header with a momentary switch on it, or two if you want both power and reset.
Click to expand...
Have open test bench and I test heavily before installing hardware in a case. I'd rather have buttons than an external power/ reset switch where I still need to fiddle with the tiny pins.
 
Teenyman45 said:
But can you explain the deal with the power and reset buttons?! I'm sick of building pcs on motherboards that lack them.
Click to expand...
Why? If you connect your front panel connectors correctly you shouldn't need them. Even if you build the PC on an open air test bench first for test purposes, you can literally start the system with a flat head screwdriver. There is no need to use onboard power and reset buttons. With manufacturing costs going up and the fact that these buttons aren't used outside of competitive benchmarking and by reviewers, it just isn't a big value add for most people. As a reviewer, I tend to use these more often than most people but I'm not hung up on them. It takes just an extra second to start a system without them.
learners permit said:
Same here I always open bench test hardware so trouble shooting if necessary is far simpler.
Click to expand...
I think a lot of people do this. However, I don't. If I'm building simple AIO or air cooled machines, I tend to build in the case. However, I'm very slow and deliberate about everything I do. I've rarely ever had machines that failed to start the first time out of hundreds of builds. Again, swapping hardware in my personal machines or on the test bench is something I do with far more frequency than most. That said, if I am doing custom water cooling with hard tubing especially, I do test on the bench first. The last thing you want to do is get some custom tubing bent and put everything in place, leak test and find out that the system won't POST because the board is DOA or whatever.
Teenyman45 said:
Have open test bench and I test heavily before installing hardware in a case. I'd rather have buttons than an external power/ reset switch where I still need to fiddle with the tiny pins.
Click to expand...
I've probably reviewed around 200 motherboards at this point. I couldn't tell you how many systems I've built or repaired. It's well into the tens of thousands I'm sure. I don't bother testing heavily outside of review hardware where that's part of the job. The fact is, if you are going to have problems they'll most likely happen right away. A decent motherboard will run with most RAM without doing anything special. It's only shitty motherboards that have to be tuned to run right. Good ones will run perfectly fine using automatic settings. I'm not saying that this is optimal, just that it will work. If it doesn't, you've probably got a hardware issue. Obviously, due diligence on BIOS updates and messing with RAM settings may be in order, but beyond that I think 24-48 hour burn in testing on the bench ahead of time is a massive waste of time.

Again, I think it depends heavily on the risk level. If you build a lot of machines, the extra caution probably isn't warranted. If I build a machine and the board is DOA, ripping it back out isn't that big a deal. Again, if custom water cooling is in the picture then it's wise to test a bit longer and test on the bench ahead of time. I would also advise testing a high end GPU vigorously before ripping off its heat sink and fans to go with a waterblock.
 
Dan_D said:
Why? If you connect your front panel connectors correctly you shouldn't need them. Even if you build the PC on an open air test bench first for test purposes, you can literally start the system with a flat head screwdriver. There is no need to use onboard power and reset buttons. With manufacturing costs going up and the fact that these buttons aren't used outside of competitive benchmarking and by reviewers, it just isn't a big value add for most people. As a reviewer, I tend to use these more often than most people but I'm not hung up on them. It takes just an extra second to start a system without them.

I think a lot of people do this. However, I don't. If I'm building simple AIO or air cooled machines, I tend to build in the case. However, I'm very slow and deliberate about everything I do. I've rarely ever had machines that failed to start the first time out of hundreds of builds. Again, swapping hardware in my personal machines or on the test bench is something I do with far more frequency than most. That said, if I am doing custom water cooling with hard tubing especially, I do test on the bench first. The last thing you want to do is get some custom tubing bent and put everything in place, leak test and find out that the system won't POST because the board is DOA or whatever.

I've probably reviewed around 200 motherboards at this point. I couldn't tell you how many systems I've built or repaired. It's well into the tens of thousands I'm sure. I don't bother testing heavily outside of review hardware where that's part of the job. The fact is, if you are going to have problems they'll most likely happen right away. A decent motherboard will run with most RAM without doing anything special. It's only shitty motherboards that have to be tuned to run right. Good ones will run perfectly fine using automatic settings. I'm not saying that this is optimal, just that it will work. If it doesn't, you've probably got a hardware issue. Obviously, due diligence on BIOS updates and messing with RAM settings may be in order, but beyond that I think 24-48 hour burn in testing on the bench ahead of time is a massive waste of time.

Again, I think it depends heavily on the risk level. If you build a lot of machines, the extra caution probably isn't warranted. If I build a machine and the board is DOA, ripping it back out isn't that big a deal. Again, if custom water cooling is in the picture then it's wise to test a bit longer and test on the bench ahead of time. I would also advise testing a high end GPU vigorously before ripping off its heat sink and fans to go with a waterblock.
Click to expand...
My front panel cables are in the middle of my TH10 case, and while that does mean they are fairly long, does not mean they reach all the way depending on how I want to orient my bench.

I water cool, so I work on the bench first. Gotta love how the CaseLabs motherboard tray doubles as a test bench. Still get a LOT of use out of my NeXXXos Monsta 4x120x80 thick radiators. With the exception of my pre-waterblocked 6900XT Speedster Zero there was days of testing before blocking a card, not just for voltage and frequency but coil whine. Oh the joys of staring at a furmark donut. DDR4 gains aren't the same as pushing DDR3 so I no longer run 24 hour memtest and prime 95 stability tests per setting. I don't build systems often, but when I do they work quite well as even a decade later one Swiftech pump is still running. Just ignore my history with Corsair parts or the hours lost when Win 10 forced updates (despite being a Pro retail license) cause settings to go haywire.
 
You must log in or register to reply here.
Back
Top