Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/nvidia-halts-testing-of-18a-process.24259/
        )

    [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
)

Nvidia halts testing of 18A process

P

prime007

Active member
Happy holidays, everyone!

It's been a while but wanted to contribute something small to this forum I read practically everyday! 😅

Intel Corporation stock dropped 2% Wednesday after Reuters reported that Nvidia stopped testing Intel’s 18A chip manufacturing process. The stock opened as much as 3.9% lower before erasing some losses.

Accordingly, Nvidia had been evaluating whether it could manufacture its chips using Intel’s advanced 18A production technology but decided not to move forward with the tests.


Related articles below:
 
Last edited by a moderator:
prime007 said:
Happy holidays, everyone!

It's been a while but wanted to contribute something small to this forum I read practically everyday! 😅

It appears Nvidia will not move forward with Intel's 18A process node.

Related articles below:
Click to expand...
Well Intel themselves said 18A will be internal only node
 
siliconbruh999 said:
Well Intel themselves said 18A will be internal only node
Click to expand...
the revisionist history is quite stunning here.

2021: 18A was the external version of the Internal node. 50 products being testing. Everyone is looking at it. Whales and prepayments coming in. 5B in external foundry revenue in 2026. Fabs planned everywhere

2024: Internal version (20A) is cancelled and product moved to TSMC. This is because 18A was doing so well, there was no need for 20A.

1H2025: No one (except USG with 2030 volume) external has signed up. LBT says maybe the external foundry fanstasy needs to stop

2H2025: Apple, Google, Nvidia Looking at it. 18A will win.. ... Risk production started in April. Panther lake Launch in 2025 Yay!

18A is "internal only" because no one external wants to use it as of today. Its pretty simple.

Everyone looks at options, then they make decisions and have backup plans based on reality.

Perhaps we should track external tapeouts for products and external revenue instead of these "news articles"

We know there is no significant (defined as greater than half a fab) external foundry volume until at least 2028.
 
MKWVentures said:
2024: Internal version (20A) is cancelled and product moved to TSMC. This is because 18A was doing so well, there was no need for 20A.

We know there is no significant (defined as greater than half a fab) external foundry volume until at least 2028.
Click to expand...
I've said it before and I'll say it again. There was a need for 20A. It was a necessary step in the development of 18A. Even in Gelsinger's Pollyanna world there were only going to be a few SKU's run on 20A. In fact, what you are calling revisionist history is really only a replacement of Gelsinger's reality with the one the rest of us live in.

One of the biggest issues with 18A is that it was still designed as an Intel first node. It is my understanding that there are still many issues with the transistor library that make it less desirable for others to use it.

As to the 2028 claim, I'm assuming you are basing that on potential 14A adoption?
 
Artificer60 said:
I've said it before and I'll say it again. There was a need for 20A. It was a necessary step in the development of 18A. Even in Gelsinger's Pollyanna world there were only going to be a few SKU's run on 20A. In fact, what you are calling revisionist history is really only a replacement of Gelsinger's reality with the one the rest of us live in.
Click to expand...
Great Comment. I agree on need for 20A. And reasons for not ramping it

Artificer60 said:
One of the biggest issues with 18A is that it was still designed as an Intel first node. It is my understanding that there are still many issues with the transistor library that make it less desirable for others to use it.
Click to expand...
I just respond to what the IFS people said to us. I think @Scotten Jones had some comments on the whole BSPD as well. If 18A was not great for external, then those "we were misled" lawsuits were correct.
Artificer60 said:
As to the 2028 claim, I'm assuming you are basing that on potential 14A adoption?
Click to expand...
2028 is not likely to be 14A volume. 2028 18A assuming that companies have started designing today on it and that Intel demonstrates 18A capability and stability. I am not sure how anything could be earlier than that. The finances on 18A are a serious challenge.

Again if we just focus on reality (tapeouts and revenue), I think we will be better off. Intel knows what is actually happening .... lets see what they tell us.
 
prime007 said:
Happy holidays, everyone!

It's been a while but wanted to contribute something small to this forum I read practically everyday! 😅

Intel Corporation stock dropped 2% Wednesday after Reuters reported that Nvidia stopped testing Intel’s 18A chip manufacturing process. The stock opened as much as 3.9% lower before erasing some losses.
Click to expand...

Happy Holidays! It really was a great one this year. Food, family, friends, boating, definitely one of the better holidays!

Nvidia was never on the intel 18A customer short list as far as I know. Nvidia and AMD are so tight with TSMC I don't see it ever happening for CPU/GPUs and other big chips. Same with AMD. Jensen Huang, Lisa Su, and CC Wei are very close friends and compatriots. TSMC N2 has a record amount of tape-outs coming, even more than N3 if you can believe that. This all points to a stronger NOT TSMC market in the years to come.

Intel 18A will have external foundry customers in 2026. I know of 4, hopefully there will be more. These will be in HVM in the 2028-2030 time frame. This update is from IEDM in early December. Hopefully none will drop out but anything can happen before taping-out. We should hear more on the next investor call. I really am expecting good things from Lip-Bu and Intel in 2026.

Samsung 2nm will have some customers as well but nothing earth shattering. Elon Musk is using both TSMC N2 and Samsung 2nm but my guess is that the bulk of the volume will go to TSMC. I also believe Elon will do a deal with Intel Foundry of some sort. We shall see. QCOM will make some Samsung 2nm chips for Samsung but most of the wafer volume will be TSMC N3 and N2 moving forward.

Bottom line: The NOT TSMC business is alive and well. Intel Foundry will lead this market segment with 18A and 14A. Just my opinion of course.
 
You must log in or register to reply here.
Back
Top