TSMC has Seven Major Customers Lined Up for its 3 nm Node

[TheLostSwede]

Based on media reports out of Taiwan, TSMC seems to have plenty of customers lined up for its 3 nm node, with Apple being the first customer out the gates when production starts sometime next month. However, TSMC is only expected to start the production with a mere 1,000 wafer starts a month, which seems like a very low figure, especially as this is said to remain unchanged through all of Q4. On the plus side, yields are expected to be better than the initial 5 nm node yields. Full-on mass production for the 3 nm node isn't expected to happen until the second half of 2023 and TSMC will also kick off its N3E node sometime in 2023.

Apart from Apple, major customers for the 3 nm node include AMD, Broadcom, Intel, MediaTek, NVIDIA and Qualcomm. Contrary to earlier reports by TrendForce, it appears that TSMC will continue its rollout of the 3 nm node as previously planned. Apple is expected to produce the A17 smartphone and tablet SoC, as well as advanced versions of the M2, as well as the M3 laptop and desktop processors on the 3 nm node. Intel is still said to be producing its graphics chiplets with TSMC, with the potential for GPU and FPGA products in the future. There's no word on what the other customers are planning to produce on the 3 nm node, but MediaTek and Qualcomm are obviously looking at using the node for future smartphone and tablet SoCs, with AMD and NVIDIA most likely aiming for upcoming GPUs and Broadcom for some kind of HPC related hardware.



View at TechPowerUp Main Site | Source

 

[bonehead123]

I'm soooo surprised....

Said absolutely N*O*B*O*D*Y

Good yields, bad yields, no yields... it won't matta.... the minute the fruity boys get their grubby paws on these nex-gen chips, EVERYBODY else will be barking & howling so loudly at TSMC, that they will have to ramp up production way faster than they originally planned to, hopefully they will be prepared...

 

[bug]

Yeah, I was really worried they'd be all dressed up with nowhere to go

 

[trsttte]

I'm soooo surprised....

Said absolutely N*O*B*O*D*Y

Good yields, bad yields, no yields... it won't matta.... the minute the fruity boys get their grubby paws on these nex-gen chips, EVERYBODY else will be barking & howling so loudly at TSMC, that they will have to ramp up production way faster than they originally planned to, hopefully they will be prepared...

Yes but let's see how things pan out with the upcoming almost certain global recession (europe at least is pretty much guaranteed). We're already seing slowdowns on some sectors. There's also the PRC headache and they might need a big distraction from their crumbling housing market and economy soon, a war usually does it.

 

[Space Lynx]

Yes but let's see how things pan out with the upcoming almost certain global recession (europe at least is pretty much guaranteed). We're already seing slowdowns on some sectors. There's also the PRC headache and they might need a big distraction from their crumbling housing market and economy soon, a war usually does it.

Inflation isn't done yet. I know several people here in England on the brink of panic because of energy bill increases, and more increases coming. That means they won't be spending as much elsewhere, the story is the same story regardless of what country you are in.

I also think this will be the year people start to get sick of yearly upgrades on phones (not the majority of that crowd, but a bigger chunk than expected) which will lead to extra free time for the TSMC factories to allocate to another company and so on and so forth.

 

[trsttte]

Inflation isn't done yet. I know several people here in England on the brink of panic because of energy bill increases, and more increases coming. That means they won't be spending as much elsewhere, the story is the same story regardless of what country you are in.

I also think this will be the year people start to get sick of yearly upgrades on phones (not the majority of that crowd, but a bigger chunk than expected) which will lead to extra free time for the TSMC factories to allocate to another company and so on and so forth.

Europe (England included) is particularly screwed because of energy, otherwise I doubt we would feel as much economic pain (inflation not counting for energy is a lot more contained than say in the US). That won't change until the war in Ukraine ends and/or there are some changes in Russia.

But Apple at least seems confident, their sales prediction and production targets for the next iPhone didn't change, I really don't understand how they can sell so many of the same small incremental update year after year.

 

[phanbuey]

Europe (England included) is particularly screwed because of energy, otherwise I doubt we would feel as much economic pain (inflation not counting for energy is a lot more contained than say in the US). That won't change until the war in Ukraine ends and/or there are some changes in Russia.

But Apple at least seems confident, their sales prediction and production targets for the next iPhone didn't change, I really don't understand how they can sell so many of the same small incremental update year after year.

Europe's/euros core problem regarding inflation is the economic policy that can control it doesn't work for the euro as a whole (i.e. what works for Germany, doesn't work for Spain/Greece etc.). Some countries need to hike rates and slow demand, and others would collapse if that happened. It's going to be interesting to see what the bank can actually do in that structure.

We are nowhere near done. We are also about to feel the food crunch once we burn through last year's reserves.

Also TSMC actually has at least 8 major customers for its 3nm node, in addition to the ones mentioned above .

 

[dicobalt]

Small number good. Big number bad. Me want small number. Oooooga Boooogga

 

[InVasMani]

Must be nice to get $20 billion dollars to dump into TSMC's 3nm. At least it'll have glued together DX9 support and offer mid range GPU performance maybe.

 

[medi01]

So, TSMC is effectively a monopoly now, as far as cutting-edge fab process goes.

It sucks...

Europe's/euros core problem regarding inflation is the economic policy that can control it doesn't work for the euro as a whole (i.e. what works for Germany, doesn't work for Spain/Greece etc.). Some countries need to hike rates and slow demand, and others would collapse if that happened. It's going to be interesting to see what the bank can actually do in that structure.

Central bank can control inflation caused by abundance of buying power.

This inflation (as in "things are getting more expensive"), however, is caused by spikes of energy sector. It is objectively more expensive to produce certain stuff.

UK inflation is at 10%. That is roughly entire Europe, give or take couple of %, is. Oh, and USA too.

Last time I've checked, neither US nor UK were in Euro (currency) zone.

 

[phanbuey]

So, TSMC is effectively a monopoly now, as far as cutting-edge fab process goes.

It sucks...


Central bank can control inflation caused by abundance of buying power.

This inflation (as in "things are getting more expensive"), however, is caused by spikes of energy sector. It is objectively more expensive to produce certain stuff.

UK inflation is at 10%. That is roughly entire Europe, give or take couple of %, is. Oh, and USA too.

Last time I've checked, neither US nor UK were in Euro (currency) zone.

The inflation happening is global - UK and US have ability to react. Again, will be interesting to see, but the traditional “tools” to manage inflation are definitely harder for the eurozone.

This inflation is a combination of a bunch of things but the traditional levers are still the same.

 

[dicobalt]

How to make sure there's never enough capacity so prices never go down.

 

[bug]

How to make sure there's never enough capacity so prices never go down.

Because traditionally, cutting edge technology is both affordable and ubiquitous, right?

 

[dicobalt]

Because traditionally, cutting edge technology is both affordable and ubiquitous, right?

Traditionally a process becomes mature then goes to production as reasonably affordable. Right now everyone is rabbit humping the lowest node number they can get, and capacity can never be enough. It's not tick tock, it's tick tick tock. Securing that node is also a very expensive business that locks out competition. Then price anchoring happens as it is now. TSMC and all their large customers have every incentive to continue this.

 

[hat]

Traditionally a process becomes mature then goes to production as reasonably affordable. Right now everyone is rabbit humping the lowest node number they can get, and capacity can never be enough. It's not tick tock, it's tick tick tock. Securing that node is also a very expensive business that locks out competition. Then price anchoring happens as it is now. TSMC and all their large customers have every incentive to continue this.

Yeah, I don't remember node advancements being this frequent in earlier years. Sadly it seems like everyone is relying on node advancements to make a better product, rather than improving the architecture itself, and then maybe the next generation might benefit from a die shrink...

 

[phanbuey]

X3D was a non-node shrink performance improvement- and a massive one at that. Sad they didn’t implement it across the board on the high end skus

 

[bug]

Traditionally a process becomes mature then goes to production as reasonably affordable. Right now everyone is rabbit humping the lowest node number they can get, and capacity can never be enough. It's not tick tock, it's tick tick tock. Securing that node is also a very expensive business that locks out competition. Then price anchoring happens as it is now. TSMC and all their large customers have every incentive to continue this.

There is truth to that, but I don't think that's the whole picture. Node shifts happened less frequently, yes, but the products were not actually cheaper. For example, a Pentium 4 EE was launched at $999 back in 2004. That's about $1,500 in today's money. More expensive than a 5950X or a 12900K.
Also, this mobile industry that wants to build a billion device for you to buy each year, if possible, wasn't around to put pressure on fabs.