News TSMC Will Reportedly Charge $20,000 Per 3nm Wafer

[Admin]

A report says TSMC will hike the price of leading-edge production at N3 by 25% compared to N5. This will impact prices on future products made using this process node.

TSMC Will Reportedly Charge $20,000 Per 3nm Wafer : Read more

 

[TechieTwo]

This is called "telegraphing". It's a means to condition consumers for significant price hikes on coming products.

 

[logainofhades]

As long as yields are good, you get more chips per wafer, on a smaller node, which should help alleviate some of the brunt of this price increase.

 

[elforeign]

Pretty soon modern electronic devices are going to become extremely cost prohibitive for the mainstream individual. More and more the technologies used to power our devices, being manufactured on the cutting edge, are just not going to be affordable.

I don't really see how we will get costs under control given the extremely low tolerances for defects in many of the production steps for semiconductors, on top of the extremely low tolerances for impurities in everything from the ultra-pure water to the etching and deposition, and overall in the sourcing of the rare-earths and raw materials that compose the lithography machinery and the end products.

 

[PlaneInTheSky]

Taiwan's monopoly on chips, motherboards and many other electronics is bad for everyone. It's bad for human development worldwide.

 

[logainofhades]

Taiwan's monopoly on chips, motherboards and many other electronics is bad for everyone. It's bad for human development worldwide.

It isn't a total monopoly. Intel makes most of their own chips and chipsets. I think they outsource some older nodes, but their cutting edge stuff, I believe is all them. Samsung is another player in the game, for electronics, and based in Korea, with plants all over the world. Micron is a US based company with manufacturing sites all over the world. Texas Instruments is another US Based company. SK-Hynix is Korean with fabs in multiple countries as well.

 

[PiranhaTech]

AMD's chiplet strategy is making a lot of sense. Smaller, simpler chiplets for the new stuff, plus some things like the I/O being on a more solid, older node just makes sense

GamersNexus had an interview with an AMD engineer that was really good

 

[Deleted member 431422]

So much for the promise of cost reduction on EUV.

 

[kjfatl]

Pretty soon modern electronic devices are going to become extremely cost prohibitive for the mainstream individual. More and more the technologies used to power our devices, being manufactured on the cutting edge, are just not going to be affordable.

I don't really see how we will get costs under control given the extremely low tolerances for defects in many of the production steps for semiconductors, on top of the extremely low tolerances for impurities in everything from the ultra-pure water to the etching and deposition, and overall in the sourcing of the rare-earths and raw materials that compose the lithography machinery and the end products.

There was a time, not too long ago where electronic products had long lifecycles. When I was a young pup, my manager told me a story about when he first started work. He was hired in the 1950's and his first project was an electronic multiplier for a business machine. Vacuum tube electronics. The product worked well, but the company put it on the shelf for 10 years. In the early 1960's the company pulled it off the shelf and sold it for over 10 years. We are headed in the same direction with IC's. The Intel I-19 and the Ryzen 17000 may have 20 year lifetimes. There will still be a place for exotic high end machines, but their use will be limited. Intel's tic-toc might still happen, with a new rev every 5 years.

 

[Deleted member 431422]

There was a time, not too long ago where electronic products had long lifecycles. When I was a young pup, my manager told me a story about when he first started work. He was hired in the 1950's and his first project was an electronic multiplier for a business machine. Vacuum tube electronics. The product worked well, but the company put it on the shelf for 10 years. In the early 1960's the company pulled it off the shelf and sold it for over 10 years. We are headed in the same direction with IC's. The Intel I-19 and the Ryzen 17000 may have 20 year lifetimes. There will still be a place for exotic high end machines, but their use will be limited. Intel's tic-toc might still happen, with a new rev every 5 years.

I wouldn't mind that to be honest. It's a race who's the fastest no matter what now. They could, I mean AMD, Intel, Nvidia take their time and refine their technologies.
I read once, people did things with Amiga that it wasn't designed for. Simply because it was as it is. No option to upgrade. Now they just slam it with more compute power and don't care about the rest.

 

[kjfatl]

I suspect that gaming as a major technology driver is over. The next big thing is probably self driving vehicles. This market will be quite different though with 50 or fewer real customers like GM and Toyota choosing what is 'good enough' for most vehicles. The hard part for vehicles is keeping the compute costs under $5000/vehicle and still getting the job done. On the other side, repurposing the powerful vehicle computers as a gaming machines could be quite interesting.

 

[-Fran-]

$2000 RTX 5080, here we go! (yes, not a theoretical 5090; Jensen needs his leather jackets)

Regards

 

[Mattzun]

Not seeing a problem here - the N3 node is still more cost efficient and better performing than N5.
Wish I could say the same thing about GPUs

If the price goes from 16000 to 20000 (1.25x) and the density is 1.7 time higher, chips built on n3 are cheaper than the same chip would be on n5 (and faster or use less power)

 

[Alvar "Miles" Udell]

Guess this means more people will use the things they have until they die or manufacturers are forced to reduce prices in order to shift inventory.

 

[Deleted member 14196]

Yes. Vote with your dollars. You don’t always need the latest or best

be Socratic. Moderation in all things

 

[Deleted member 14196]

$2000 RTX 5080, here we go! (yes, not a theoretical 5090; Jensen needs his leather jackets)

Regards

He should get a bad arse duster too

 

[spongiemaster]

As long as yields are good, you get more chips per wafer, on a smaller node, which should help alleviate some of the brunt of this price increase.

That doesn't apply to GPU's as Nvidia and AMD stick to roughly the same die sizes from gen to gen.

 

[JayNor]

Intel's largest MAX GPUs have 16 compute tiles using TSM N5. The successor, Rialto, is on the roadmap with 20 compute tiles.

 

[fbradcdsc]

Again,the 16k price for a 5nm wafer is from 2020.
And 20k is actually pretty amazing considering costs per wafer used to basically double gen on gen.
As for the yields we'll know soon enough

 

[RichardtST]

I think we've hit a critical point in the "law of diminishing returns". I'm simply not willing to pay more for my electronics. Forget it. This is why I'm not excited about the new AMD processes either. Double the price for a scant 15% improvement is a no-go. You double the price, I want more than double the power. Big tech can bite me. I'm going low tech.

 

[Ogotai]

This is why I'm not excited about the new AMD processes either. Double the price for a scant 15% improvement is a no-go

ahh so you were happy paying intel for its less then 5% performance increase ( maybe 10% in some cases ) gen on gen, while they kept raising prices, while keeping the main stream at quad cores, and making you pay quite a bit more for 6+ cores on their HEDT platform, then ?

 

[spongiemaster]

ahh so you were happy paying intel for its less then 5% performance increase ( maybe 10% in some cases ) gen on gen, while they kept raising prices, while keeping the main stream at quad cores, and making you pay quite a bit more for 6+ cores on their HEDT platform, then ?

He didn't say that, and what you said is false as well.

Sandy Bridge 2700k - released Jan 2011 - $332
Kaby Lake 7700k - released Jan 2017 - $339

6 years, 6 releases, price increased by a total of $7. 2% price increase while inflation was 9% over the same time period.

 

[Ogotai]

He didn't say that, and what you said is false as well.

i know he didnt say that, i was asking him if he was happy with intel charging more each gen for 5% performance increases

Sandy Bridge 2700k - released Jan 2011 - $332
Kaby Lake 7700k - released Jan 2017 - $339

6 years, 6 releases, price increased by a total of $7. 2% price increase while inflation was 9% over the same time period.

maybe where YOU were prices didnt increase that much, but here, they did, and those increases were more then a few bucks each gen, the increases where quite noticeable, and more then they should of been. when Zen 1 was release, intels top end, dropped in price by 1k, the rest of the stack, dropped a few hundred for each tier. i know quite a few people who bought 2700k and 2600k's back then, and most of them were STILL using those same cpus up til about 2 years ago, when they finally upgraded, and they upgraded to AMD, NOT intel, one friend is still using his 2600k to this day, they all had the same reason, intels performance increases gen on gen were NOT worth the money intel was charging here. they switched to AMD with Zen 2 and zen 3, and the friend that still using the 2600k will be picking up Zen 4 early next year.

so sorry spongiemaster, but i was not wrong, or false. but feel feel to think i was.

 

[RichardtST]

ahh so you were happy paying intel for its less then 5% performance increase ( maybe 10% in some cases ) gen on gen, while they kept raising prices, while keeping the main stream at quad cores, and making you pay quite a bit more for 6+ cores on their HEDT platform, then ?

No, of course not. I've still got my old 2012-ish i7-4770. Kept that little rascal active for nearly 10 years. Now it is semi-retired just running VR in the living room. I'm actually quite annoyed that processors in general only look like they go faster because of all the bling... Storage however (NVME), has been a recent blessing in speed!

 

[bit_user]

Pretty soon modern electronic devices are going to become extremely cost prohibitive for the mainstream individual.

That's only true if the chips stay the same size. However, while the new wafers are getting more expensive, they're also denser. This enables existing chips to be fabbed on the new nodes at the same or lower prices. The main way that chips are getting more expensive is they're increasing in complexity enough to nullify the density-adusted cost savings, and then some. Plus, there are various non-recurring costs, which are also going up.

What I'd have really liked to see is a table of the per-transistor cost, for all the process nodes. It was awesome to see the wafer prices all listed, but that doesn't tell us the whole story.

I don't really see how we will get costs under control given the extremely low tolerances for defects in many of the production steps for semiconductors,

Chiplets help counter the issue of defects.