Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/who-is-the-intel-18a-large-customer-prepay.18690/
        )

    [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] => 2021370
            [XFI] => 1050270
        )

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

Who is the Intel 18A Large Customer Prepay?

Daniel Nenni

Daniel Nenni

Admin
Staff member
1693674170369.png


"And we've now received a large customer prepay for 18A capacity. So now customers are getting confident enough that they're putting dollars on our balance sheet to accelerate our 18A capacity. So quite excited about that." Pat Gelsinger 8/31/2023.

Let's make a list of potential Intel 18A prepay customers:

AMD
Apple
Broadcom
Huawei
Intel
Marvell
MediaTek
NVIDIA
QCOM
Samsung VLSI

Anybody else?

And if you had to pick one who would it be?
 
Google and AWS.
Google for their TPU and IPU
And AWS make want to do custom x86 designs with Intel to take control of design and be a bit more self reliant and control costs.
 
Daniel Nenni said:
I needed ten for the list. It is not impossible though but Huawei would not be on my list. I only see three possible candidates.
Click to expand...
Which three? I see NVDA, QCOM and maybe Apple.

Why Apple? 1. The agreement to not disclose the customer, Apple tend to hide its secret until it no longer is. 2. Apple may want to introduce some optionality, like how they previously welcomed Intel to serve modem components until Intel failed and sold that unit to Apple. They can potentially work again because it’s not a direct conflict of interest. People who buy Mac may still need x86 for its backward compatibility or customization it can provide. And vice versa. Intel and AMD are directly competing with another. Not so much between Intel and Apple. 3. We haven’t see chiplet push by Apple yet, but it doesn’t mean that they aren’t doing it behind the scene. The number of breakthrough they can do with chiplet is considerably large. Like M1 Max and M1 Ultra, maybe they can do something differently that reduce the area and cost. Various technologies from different companies will certainly expand the toolbox and so does the supply.
 
delong.height said:
Google and AWS.
Google for their TPU and IPU
And AWS make want to do custom x86 designs with Intel to take control of design and be a bit more self reliant and control costs.
Click to expand...
AWS does not appear to want to make custom x86 designs. AWS uses Arm core IP (and perhaps an Arm architecture license) to design their own CPUs, called Gravitons, and they are already widely deployed in EC2 and as application servers for various AWS applications, including Redshift and Aurora, to name just two of many.

aws.amazon.com

ARM Processor - Performance Processor - AWS EC2 Graviton - AWS

AWS Graviton2 enables the best price/performance for your cloud workloads.
aws.amazon.com aws.amazon.com
 
Probably Nvidia because they can use whatever capacity they can get for AI over the next 3-5 years..
 
PDK 0.9 for 18A coming soon.



Also, from the interview CFO Dave had:



Dave Zinsner

Yes. I would characterize it as -- and we're engaged with several customers. And as we talked about, we thought we could get what we call a whale done…

Chris Danely

Will this be a whale?

Dave Zinsner

By the end of the year, this will be a whale, by the end of the year on 18A.




So, it's not Nvidia or Qualcomm (given he said "by the end of the year, this will be a whale), I guess someone like Ampere or Marvell.
 
delong.height said:
PDK 0.9 for 18A coming soon.
Click to expand...
delong.height said:
Also, from the interview CFO Dave had:
Dave Zinsner
Yes. I would characterize it as -- and we're engaged with several customers. And as we talked about, we thought we could get what we call a whale done…

Chris Danely
Will this be a whale?

Dave Zinsner
By the end of the year, this will be a whale, by the end of the year on 18A.

So, it's not Nvidia or Qualcomm (given he said "by the end of the year, this will be a whale), I guess someone like Ampere or Marvell.
Click to expand...

I believe you are quoting from a talk at the 2023 Citi Global Technology Conference that Intel CFO David Zinsner attended. The audio webcast replay can be found here:

www.intc.com

Citi Global Technology Conference

www.intc.com www.intc.com
 
I would say Google or Amazon (whale company). Maybe second source for Marvell or broadcom. Was the contract signed and a payment made to Intel already?
 
MKWVentures said:
I would say Google or Amazon (whale company). Maybe second source for Marvell or broadcom. Was the contract signed and a payment made to Intel already?
Click to expand...

It seems to be a well kept secret. My guess would be a high performance type application versus an SoC or anything power sensitive.

Hopefully we can get more details on what prepay means since there is no standard definition for foundries that I have seen.

Not sure what a semiconductor whale is either. I guess it could be a huge company with small volumes or a company with huge volumes?
 
delong.height said:
Which three? I see NVDA, QCOM and maybe Apple.

Why Apple? 1. The agreement to not disclose the customer, Apple tend to hide its secret until it no longer is. 2. Apple may want to introduce some optionality, like how they previously welcomed Intel to serve modem components until Intel failed and sold that unit to Apple. They can potentially work again because it’s not a direct conflict of interest. People who buy Mac may still need x86 for its backward compatibility or customization it can provide. And vice versa. Intel and AMD are directly competing with another. Not so much between Intel and Apple. 3. We haven’t see chiplet push by Apple yet, but it doesn’t mean that they aren’t doing it behind the scene. The number of breakthrough they can do with chiplet is considerably large. Like M1 Max and M1 Ultra, maybe they can do something differently that reduce the area and cost. Various technologies from different companies will certainly expand the toolbox and so does the supply.
Click to expand...

AMD
Apple
Broadcom
Huawei
Intel
Marvell
MediaTek
NVIDIA
QCOM
Samsung VLSI

It could certainly be a chiplet deal where a CPU/GPU chiplet is made at Intel and the other chiplets are made at TSMC.

Not Apple for sure, or AMD, or Huawei, or Samsung or Intel. Intel calling Intel a whale customer would be funny but not funny. Someone mentioned Ericcson 5G chips, a definite possibility, that is a whale of a company but not a whale of volume.

I hope it is Nvidia.
 
Daniel Nenni said:
It seems to be a well kept secret. My guess would be a high performance type application versus an SoC or anything power sensitive.

Hopefully we can get more details on what prepay means since there is no standard definition for foundries that I have seen.

Not sure what a semiconductor whale is either. I guess it could be a huge company with small volumes or a company with huge volumes?
Click to expand...

I'm not sure why Intel CFO David Zinsner called this potential and important IFS customer a "whale". A whale in financial, investment, and banking industry may not be a good creature. For example: the "London Whale" - it costed JPMorgan Chase whopping US$7+ billion loss and fines.

2012 JPMorgan Chase trading loss - Wikipedia

en.wikipedia.org en.wikipedia.org
 
You must log in or register to reply here.
Back
Top