News Intel to fab chips for Qualcomm

[NTMBK]

Intel’s First High-Profile IFS Fab Customer: Qualcomm Jumps on Board For 20A Process

www.anandtech.com

Starting at the 20A process in a few years. I wonder what parts they will build at Intel? If it's the laptop chips meant to compete with Intel, that's going to be weird.

 

[Gideon]

I don't believe anyone shared this yet:

Intel Grossly Exaggerates Foundry “Wins” With Qualcomm And Amazon, There Are No Deals For Manufacturing Chips

Intel touted huge wins with Qualcomm and Amazon for their foundry business, but a deeper investigation reveals there are no chips deals penned. Intel will only package for Amazon, and Qualcomm is only evaluating the technology.

semianalysis.substack.com

1. Amazon "win" is just for packaging, chips will still be made at TSMC
2. Qualcomm has 0 product plans for Intel yet, they are "just evaluating" (e.g. probably playing this to get better deals from Samsung and TSMC)

To call these things major customer wins ... one can only imagine the quality of those touted 100+ potential customers. Did Intel hand out fliers that give you a free lunch at their caffeteria when you fill a form to sign up as their future node customer?

 

[Hitman928]

I don't believe anyone shared this yet:

Intel Grossly Exaggerates Foundry “Wins” With Qualcomm And Amazon, There Are No Deals For Manufacturing Chips

Intel touted huge wins with Qualcomm and Amazon for their foundry business, but a deeper investigation reveals there are no chips deals penned. Intel will only package for Amazon, and Qualcomm is only evaluating the technology.

semianalysis.substack.com

1. Amazon "win" is just for packaging, chips will still be made at TSMC
2. Qualcomm has 0 product plans for Intel yet, they are "just evaluating" (e.g. probably playing this to get better deals from Samsung and TSMC)

To call these things major customer wins ... one can only imagine the quality of those touted 100+ potential customers. Did Intel hand out fliers that give you a free lunch at their caffeteria when you fill a form to sign up as their future node customer?

The 100 potential customers just means they showed some level interest above zero. I'm not saying there aren't any very serious opportunities for Intel in the bunch, but the number they gave is essentially meaningless at this point in time.

Edit: I do think there will be significant interest from many companies if Intel as a foundry can 'get it right', but I'm also sure those companies will do their due diligence before jumping into bed with Intel on this.

 

[Mopetar]

The 100 potential customers just means they showed some level interest above zero.

100 people who forgot to deselect the "Send me your newsletter and keep me informed about future offers" checkbox when signing up for an account.

I'm kidding, but it's a marketing piece. Were you expecting anything resembling the truth?

 

[Roland00Address]

Intel always planned their 7nm was going to be made with EUV. Do you have a source showing otherwise?

I am saying Intel 10nm was made without EUV, and that is really hard to do anything that is 90 million transistors per mm2. Intel did that but is it worth it?

Intels future node which was called 7nm, now called Intel 4, will use EUV and I never claimed they decided to not use EUV for that node. I am saying it is good they are finally using EUV for their leading node (yes it was always their intent, but we are moving from the 14nm to various 10nm transition, to the various 10nm to Intel 4 generation.)

(yes there is also an intel 7, intel 10nm is 10nm SuperFin (10SF) , and intel 7 was called 10nm Enhanced Super Fin or 10ESF but it got renamed. Not higher density just other improvements on this variant intel 10nm which is still more dense than TSMC 7nm but not some of their newer TSMC geometries)

 

[maddie]

View attachment 48028

I am saying Intel 10nm was made without EUV, and that is really hard to do anything that is 90 million transistors per mm2. Intel did that but is it worth it?

Intels future node which was called 7nm, now called Intel 4, will use EUV and I never claimed they decided to not use EUV for that node. I am saying it is good they are finally using EUV for their leading node (yes it was always their intent, but we are moving from the 14nm to various 10nm transition, to the various 10nm to Intel 4 generation.)

(yes there is also an intel 7, intel 10nm is 10nm SuperFin (10SF) , and intel 7 was called 10nm Enhanced Super Fin or 10ESF but it got renamed. Not higher density just other improvements on this variant intel 10nm which is still more dense than TSMC 7nm but not some of their newer TSMC geometries)

TSMC 7nm was also without EUV and it definitely seems to have been worth it.

 

[Doug S]

TSMC 7nm was also without EUV and it definitely seems to have been worth it.

Good point.

Especially since Samsung's 7nm, which did use EUV, took longer to get to mass production. By using EUV for "N7+", but not for N7 or N7P, TSMC was able to take a bit more time to get up to speed with EUV mass production without jeopardizing the timeline for their biggest customer.

If Intel had added a few EUV layers in one of 10nm's "+" cycles they could have eased into it - and perhaps even solved some of the issues that have made 10nm such a problem for them to fully roll out. Their 7nm / Intel 4 process going all-in on EUV may not be so smooth since they lack the several years of mass production experience with EUV that TSMC and Samsung both have.

 

[maddie]

Good point.

Especially since Samsung's 7nm, which did use EUV, took longer to get to mass production. By using EUV for "N7+", but not for N7 or N7P, TSMC was able to take a bit more time to get up to speed with EUV mass production without jeopardizing the timeline for their biggest customer.

If Intel had added a few EUV layers in one of 10nm's "+" cycles they could have eased into it - and perhaps even solved some of the issues that have made 10nm such a problem for them to fully roll out. Their 7nm / Intel 4 process going all-in on EUV may not be so smooth since they lack the several years of mass production experience with EUV that TSMC and Samsung both have.

This brings up an interesting point.

One of the reasons that Intel went for so big a jump was their fear of being challenged by TSMC. We see how that worked out. Now, they seem to be announcing many new tech leadership pathways and claim to be back ahead in 4 years.

Deja Vu all over again? I hope not.

 

[oak8292]

I think the answer the question on why Qualcomm is working with Intel is in this announcement from the Government on a contract for RAMP-C.

"This opportunity has been awarded to Qualcomm Technologies and Intel Technologies."

The Department of Defense (DoD) currently has no on-shore access to foundry technology capable of meeting the long-term leading edge (<7nm Digital Complementary Metal-Oxide Semiconductor (CMOS) node) microelectronics fabrication needs for DoD specific designs or commercial-off-the-shelf (COTS) components. Most of the leading United States (U.S.) fabless semiconductor companies manufacture their components off-shore.

Rapid Assured Microelectronics Prototypes - Commercial (RAMP-C) - NSTXL %

S²MARTS Project No. 21-07 This opportunity has been awarded to Qualcomm Technologies and Intel Technologies

nstxl.org

 

[DrMrLordX]

@oak8292

Interesting take. This way the Pentagon has dual-source on chip design with a collaborative manufacturing facility to guarantee that supply can't be interrupted in the event of foreign turmoil.