Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/tsmc-is-set-to-begin-trial-production-of-2nm-chips-next-week.20591/
        )

    [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] => 2030770
            [XFI] => 1060170
        )

    [wordpress] => /var/www/html
)

TSMC is Set to Begin Trial Production of 2nm Chips Next Week

If this is true, TSMC will be shipping 5x the capacity on N2 vs Intel 20A and 18A combined in Sept 2025.

Why do people say "TSMC has committed all N3 capacity to Apple" its obviously not true. TSMC is not turning away major customers, they have ways to deal with upsides in demand.
 
MKWVentures said:
If this is true, TSMC will be shipping 5x the capacity on N2 vs Intel 20A and 18A combined in Sept 2025.

Why do people say "TSMC has committed all N3 capacity to Apple" its obviously not true. TSMC is not turning away major customers, they have ways to deal with upsides in demand.
Click to expand...

What's supposed to happen Sep '25?
 
fansink said:
Right on time, as usual.

It's going to be an interesting race, although possibly lonely.


9to5mac.com

TSMC ahead of schedule on trial production of 2nm chips

Apple chipmaker TSMC is beginning trial production of 2nm chips, set to be used in the iPhone 17 Pro and...
9to5mac.com 9to5mac.com
Click to expand...
iPhone 17 Pro based on this will ship in Sept 2025 based on historical patterns, and Panther Lake (Intel 18A) is due in the September-October 2025 timeframe.

TSMC N2 will undoubtedly be higher volume, but 18A will have backside power delivery and presumably noteably higher transistor performance. Panther Lake *appears* to be mobile though (due to “Arrow Lake refresh”), so ‘full clock desktop’ probably not until 2H 26 on 18A..
 
Xebec said:
iPhone 17 Pro based on this will ship in Sept 2025 based on historical patterns, and Panther Lake (Intel 18A) is due in the September-October 2025 timeframe.

TSMC N2 will undoubtedly be higher volume, but 18A will have backside power delivery and presumably noteably higher transistor performance. Panther Lake *appears* to be mobile though (due to “Arrow Lake refresh”), so ‘full clock desktop’ probably not until 2H 26 on 18A..
Click to expand...
There will be no N2 products in 2025. TSMC claims first N2 revenue will be in 1H26. Depending on if Apple pays 0d or 30d after delivery, fastest possible case is an Mx series laptop/ipad chip around mid year in late Q2'26 (similar to M4 launch). Worst case is late Q3'26 iPhones. While N2 will surely be higher volume than 18A (especially with the disaggregation trends at intel and losing marketshare in a saturated market), but that volume crossover point will presumably not occur until early/mid 2027 when they ramp non Apple customers/Apple runs more than just their lower volume halo products on N2. The fact that this misconception still exists makes me worried that folks who are out of the know will dog pile on TSMC for being "a year late" even when they really were delivering exactly on time.

Source:
C. C. Wei
Randy, the N2's ramp profile we say is very similar to N3 because of, look at the cycle time, we start the N2 production in the second half of 2025, actually in the last quarter of 2025. And because of the cycle time and all the kind of back-end process, and so we expect the meaningful revenue will start from the end of the first quarter or beginning of the second quarter of 2026. That's what we mean that is the profile is very similar to N3.

Now your second question is there have been a lot of engagement and the tape-out will be higher, and do we see a very steep kind of a production? Well, we do expect that, but let me say again, N2 is a very complicated work or a very complex technology node. So my customer, they also take a little bit longer time to prepare for the tape-out. So that's why they all engage with TSMC in the early stage. And -- but for their product ramp-up, they will have their own product road map and their own business consideration. However, we still say that N2 will be a very, very big node for TSMC.
Click to expand...
 
Last edited:
nghanayem said:
There will be no N2 products in 2025. TSMC claims first N2 revenue will be in 1H26 depending on if Apple pays 0d or 30d after delivery, fastest possible case is an Mx series laptop/ipad chip around mid year in late Q2'26 (similar to M4 launch). Worst case is late Q3'26 iPhones. While N2 will surely be higher volume than 18A (especially with the disaggregation trends at intel and losing marketshare in a saturated market), but that volume crossover point will presumably not occur until early/mid 2027 when they ramp non Apple customers/Apple runs more than just their lower volume halo products on N2. The fact that this misconception still exists makes me worried that folks who are out of the know will dog pile on TSMC for being "a year late" even when they really were delivering exactly on time.

Source:
Click to expand...

What's the date on C. C. Wei's statement?
 
nghanayem said:
There will be no N2 products in 2025. TSMC claims first N2 revenue will be in 1H26. Depending on if Apple pays 0d or 30d after delivery, fastest possible case is an Mx series laptop/ipad chip around mid year in late Q2'26 (similar to M4 launch). Worst case is late Q3'26 iPhones. While N2 will surely be higher volume than 18A (especially with the disaggregation trends at intel and losing marketshare in a saturated market), but that volume crossover point will presumably not occur until early/mid 2027 when they ramp non Apple customers/Apple runs more than just their lower volume halo products on N2. The fact that this misconception still exists makes me worried that folks who are out of the know will dog pile on TSMC for being "a year late" even when they really were delivering exactly on time.

Source:
Click to expand...
Thank you for this - I was always a little curious about these dates. From a distance, it always seemed like Samsung announces volume manufacturing, then it takes a few years to actually ship products in volume. TSMC announces HVM and it’s usually not until 6-12 months later that we see products that are retail available (based on some google searches for N7, N5, N3 and Apple). Intel has always done something else, perhaps inbetween these, depending upon what era of Intel we’re talking about :). (Intel 3 is extra muddy because it’s not a retail launch, but more of Server OEM first).

In other manufacturing industries — automobiles, a new car plant will run a pilot line/low volume for maybe a few months, and some of those early cars actually make it to customers, before high volume is ramped up over the following year (or more). In some cases, a few cars are even sold before a manufacturer says “high volume” (i.e. Tesla). I mention this because I’m trying to calibrate my understanding of “What is HVM” in semiconductors, and it’s a bit different than some other industries.

I know Apple has to stockpile a large number of chips before they release a product, to ensure availability. That means their foundry really does need to be HVM for many months before launch. Some launches are larger than others (N3 was iPhone Pro only, while N5 was the full iPhone 12 lineup).

So I guess unfortunately we have to take the foundry’’s word on “HVM” (TSMC, Intel, Samsung) and then see what actually ships to customers up to 1 year later to see if prove their announcement date?. (I know there are financial reasons for them not to mis-state, but things can happen after HVM too.. ).
 
Hopefully, TSMC has decided to break their cadence of HVM new nodes just in time for processing Apple’s September products release.

Perfect timing for adding unanticipated pressure to competitors, while allowing more time to improve yields.

I believe Nenni said earlier that TSMC already had 99% of 2nm orders, if true, this early move may improve that 99%.
 
Xebec said:
Intel has always done something else, perhaps inbetween these, depending upon what era of Intel we’re talking about :). (Intel 3 is extra muddy because it’s not a retail launch, but more of Server OEM first).
Click to expand...
In the past it would seem like intel would sometimes talk about when HVM started but most of the time they focused on the date when product would be on the market. Now the waters are a bit muddier since intel is both a chip company and a foundry now. They talk about product release date (like a chip company would) and when the process starts HVM for the lead product (similar to terminology to TSMC).
Xebec said:
In other manufacturing industries — automobiles, a new car plant will run a pilot line/low volume for maybe a few months, and some of those early cars actually make it to customers, before high volume is ramped up over the following year (or more). In some cases, a few cars are even sold before a manufacturer says “high volume” (i.e. Tesla). I mention this because I’m trying to calibrate my understanding of “What is HVM” in semiconductors, and it’s a bit different than some other industries.

I know Apple has to stockpile a large number of chips before they release a product, to ensure availability. That means their foundry really does need to be HVM for many months before launch. Some launches are larger than others (N3 was iPhone Pro only, while N5 was the full iPhone 12 lineup).
Click to expand...
Cycle times on autos are a good bit shorter than leading edge semis. I think TSMC was talking about cycle times that are over a quarter for N2 (but I could be misremembering and I don't feel like digging for the quote). As an example let's say N2 begins risk starts for the PRQ candidate stepping @BOQ4'25. First silicon outs will be BOQ1'26. Those wafers will then get shipped out to an OSAT for die sort and dicing. From there they might go to the same OSAT or even a different one to get co-packaged with LPDDR5-X and a QCOM modem. Let's call that a month. From there we would water ship from TW to the PRC so it can be put into a phone. Let's call that a month between boat leaving TW to assembled phone on a boat leaving China. Add in a 3 months for the transpacific journey and a month of distribution to get to stores. Tallying that all up means that even if Apple wanted to just launch the very first CPU produced the earliest it could get to a bestbuy in NY would be like July'26. Obviously one phone isn't sufficient and in practice that lead cert material will be expedited so Apple can do the validation while HVM is going on. But I think it illustrates the point. Of how long fab HVM vs product in hand is.
Xebec said:
So I guess unfortunately we have to take the foundry’’s word on “HVM” (TSMC, Intel, Samsung) and then see what actually ships to customers up to 1 year later to see if prove their announcement date?. (I know there are financial reasons for them not to mis-state, but things can happen after HVM too.. ).
Click to expand...
Yup. The only way to know everything was above board is seeing the proof in the pudding a year after the announcement of HVM and or risk starts.
 
You must log in or register to reply here.
Back
Top