TSMC Struggling to Make Enough 3-Nanometer Chips for Apple

[Mr. Dee]

Then by March, we see steep discounts on the 14 inch M3 Pro MacBook Pro on Amazon.com.

 

[jimbobb24]

Or maybe it's the countdown to China invading Taiwan. 🤔

Then I hope we build some fabs in Europe and the US ASAP.

 

[Andres Cantu]

Seems like we're going to be stuck with 5nm for a while. That tech will have had quite the run.

 

[RogerWilco]

Then I hope we build some fabs in Europe and the US ASAP.

TSMC is building a $25B, 5nm facility just south of Phoenix. It won’t be online until 2025.

 

[CWallace]

Idk why Apple would waste the successful yields on A17 chips. They should be prioritizing M3 for the computers, iPhones don't need to be any faster. Hopefully they are.

Apple will likely sell more A17-powered iPhone 15 Pro and Pro Max over the opening week then they will likely sell M3-powered Macs over an entire year (even if every model was refreshed with an M3 family SoC) so it should not be surprising where Apple is focusing what production they can get.

 

[tipoo]

What happens after we reach 1 tho

Fabrication plant naming is 100% marketing fluff at this point. TSMC 7nm had 40-50nm gate pitches and fin widths. We still have plenty to shrink in three dimensions, take it from Jim Keller.

 

[SouthBronx187]

replace silicon with possible other exotic material such as carbon nanotube, with current silicon wafer, the possibilities of electrons jumping lanes increases since the material application of silicon cannot be made small enough past a certain point.

💀

 

[SpotOnT]

0.5 nanometer, 0.25 nanometer, 0.01 nanometer, 0.001 nanometer.

They really need to change these node names - this is getting ridiculous.

 

[SpotOnT]

We're at a point in die sizing that is almost impossible to perfect. Its hard to blame TSMC, when it's unlikely anyone else could meet this demand today. BTW 3nm is 12 atoms wide. I realize the wires are actually larger than 3nm, but we're honesty pushing up against the laws of physics. It's unlikely Apple will be able to release chips on a smaller process, each year. We'll be lucky to see a 3 year cadence.

A 3 nm node size has nothing to do with anything physical or real. It is just a made up name.

The gate pitch for TSMC 3 nm node is 48 nanometers. 48 nanometers can still be shrunk.

 

[tipoo]

They really need to change these node names - this is getting ridiculous.

Intel already is moving to Angstroms next

 

[emoore33]

Intel already is moving to Angstroms next

I won’t be impressed until they are down to the Planck length.

 

[Serqetry]

Apple will likely sell more A17-powered iPhone 15 Pro and Pro Max over the opening week then they will likely sell M3-powered Macs over an entire year (even if every model was refreshed with an M3 family SoC) so it should not be surprising where Apple is focusing what production they can get.

They could use the last chip in the iPhone 15 if they improved it in other ways. Better cameras, USB-C, etc. It does not need a new chip, the current one is more powerful than anyone needs.

 

[smulji]

If TSMC can’t ramp up in time iphone 15 & 15 plus may get the A16 from last yeat

The regular 15 series wasn't going to get the A17 this year so that's a given

 

[dgdosen]

TSMC is building a $25B, 5nm facility just south of Phoenix. It won’t be online until 2025.

Aren't they building a 5nm and 3nm?

 

[steve123]

This article confirms Pro Max Ultra phones and computers using N3 in 2023. We are on track for debut of the Mac Pro at WWDC!

 

[svish]

Looks like Macs will get 3nm chips only in 2024

 

[KPOM]

If you go by the same publication Intel is having to buy their 3nm stock from TSMC. Looks like Intel has already given up.

Intel Will Rely on TSMC for its Rebound - EE Times

Intel will join Apple to order the world’s first 3-nm chips from TSMC as the Taiwan chip foundry ramps up its newest process this year.

www.eetimes.com

They are focusing on the smaller nodes. Buying 3nm from TSMC is a stopgap.

 

[EntropyQ3]

Intel has started Angstroms so if TSMC follows suit, at 2nm that is 20A, then 1nm is 10A, 0.7nm is 7A, for example. Pico is the next frontier however… but at some point mankind can only shrink so far. So they will hit a wall.

😀
Marketing people have neither limits nor morals, so pico is a near certainty on the future.
Atoms however DO have sizes, typically 0.2nm or so, and you need a number of them to build something and to connect that function to its surrounding. (Preferably at reasonable cost for devices with 100 billion or so gates, labs don’t count.)
We are at the end-game of lithographic shrinking. There is enormous economic pressure to extend the cycles of improvement, but …

 

[tomekwsrod]

I'm not sure if we need a 3nm. Old chip seems to be very powerful. I'd love to see lower price, better camera and software.

 

[iBug2]

Intel has started Angstroms so if TSMC follows suit, at 2nm that is 20A, then 1nm is 10A, 0.7nm is 7A, for example. Pico is the next frontier however… but at some point mankind can only shrink so far. So they will hit a wall.

Tell that to Ant-Man.

 

[SoldOnApple]

Sub nm chips are going to offer Apple and all device makers significant marketing opportunities. It just sounds cool, even if no one understands or cares what it means, it'll feel like a new age of computing is being ushered in after the marketing department is done with it.

 

[ksec]

The media continue to make up BS about TSMC. Just sad.

 

[ghanwani]

Apple has quite the appetite for chips. Hope they work out.

 

[wanha]

What happens after we reach 1 tho

We'll keep pushing until we get to imaginary numbers, known as i.

And that's what the i in iPhone, iPad, iTunes, etc meant all along.

 

[eldho]

We'll keep pushing until we get to imaginary numbers, known as i.

And that's what the i in iPhone, iPad, iTunes, etc meant all along.

And imagine how rumours will take off once we get fully into imaginary computers