Discussion Intel Foundry

[Hermitian]

What do you folks think of Intel's Foundry subsidiary?
Additionally, in the long run will it be all that remains of the current company?

https://www.streetinsider.com/dr/news.php?id=23727093&gfv=1

 

[adroc_thurston]

What do you folks think of Intel's Foundry subsidiary?

14A is hail mary. If they nail it, they're in the game. Otherwise, gg no re

 

[gdansk]

I think the general consensus is that it's the part that won't remain...

 

[Doug S]

I think the general consensus is that it's the part that won't remain...

Between the US goverment's needs and the worries everyone has about putting all their eggs in the Taiwan basket, it will succeed. Even if TSMC truly diversified (i.e. actually had leading edge capacity in Arizona/elsewhere rather than N+1) no one wants to see a single company with a monopoly on all leading edge production for the entire world.

 

[gdansk]

Between the US goverment's needs and the worries everyone has about putting all their eggs in the Taiwan basket, it will succeed. Even if TSMC truly diversified (i.e. actually had leading edge capacity in Arizona/elsewhere rather than N+1) no one wants to see a single company with a monopoly on all leading edge production for the entire world.

Well let me rephrase it: it is the side of Intel that isn't expected to "thrive" any time soon.

 

[Hermitian]

For those who buy Supercomputers in the black, a single machine is about $300 million. If the secret procurement is from a U.S. government entity, then the manufacturer must also be in the U.S.

 

[cebri1]

14A is hail mary. If they nail it, they're in the game. Otherwise, gg no re

18A is the key. 14A is too far away into the future, that node won’t generate any positive cash until 28. If 18A is not good, RIP IFS.

 

[darkswordsman17]

Between the US goverment's needs and the worries everyone has about putting all their eggs in the Taiwan basket, it will succeed. Even if TSMC truly diversified (i.e. actually had leading edge capacity in Arizona/elsewhere rather than N+1) no one wants to see a single company with a monopoly on all leading edge production for the entire world.

Ask South Korea and Samsung if you can fix things by just having the government protect and pour in resources. There's no magic wand to making it happen.

Which, the US government has lots of need for not-bleeding edge chip production so I don't know they even care about trying to keep up with TSMC. They're only just now really even attempting to do anything about the overall situation and pretty sure they've given Intel less than Israel, and Intel is a major American company and has been struggling for like a decade at trying to stay competitive in chip fabrication. They've also let Intel sell memory and other fabs/factories to non-American companies. They didn't even attempt to help GlobalFoundries and didn't care at all when GF abandoned cutting edge development. Also they let it be purchased by foreign entities (that I think were open about them taking the tech back to the Middle East to start building chip production there as well).

Most likely what happens is the US pours money trying to get Intel to be good again, the return for that is not worth it (even if its ok), and they resort to espionage and other tactics to try to make up the gap. They being Intel first and then probably the US government if the situation continues to be a problem. Heck, Samsung already did that and still can't compete with TSMC.

 

[Hitman928]

18A is the key. 14A is too far away into the future, that node won’t generate any positive cash until 28. If 18A is not good, RIP IFS.

Intel already said that IFS as a whole won’t generate positive cash flow until 2027 at the earliest, so. . .

 

[cebri1]

Intel already said that IFS as a whole won’t generate positive cash flow until 2027 at the earliest, so. . .

And that’s going to 90% 18A. If they have to wait for 14A… they’ll need another 20B just cover opex.

 

[Hitman928]

And that’s going to 90% 18A. If they have to wait for 14A… they’ll need another 20B just cover opex.

That’s why Intel is cutting costs left, right, and center as well as halting fab build outs. They were expecting more demand than they are actually getting. Hopefully that can change with some proven products on 18a, but it’s going to be a brutal near term (at least) for IFS.

 

[Hermitian]

That’s why Intel is cutting costs left, right, and center as well as halting fab build outs. They were expecting more demand than they are actually getting. Hopefully that can change with some proven products on 18a, but it’s going to be a brutal near term (at least) for IFS.

Thank you.

 

[adroc_thurston]

18A is the key

0 customers. It's 14A or bust.
They need to execute on 18A internally and then maybe they'll get stuff lined up. Which is 14A.

 

[cebri1]

Let’s treat this as the IFS thread until the forum dies.

 

[DrMrLordX]

0 customers. It's 14A or bust.

Didn't Broadcom dog them out for 18a not meeting expectations? That might be a hint as to why they're not getting customers.

 

[Hermitian]

0 customers.

... that the public is aware of.

 

[adroc_thurston]

Didn't Broadcom dog them out for 18a not meeting expectations?

Broadcom stuff was whatever. Qualcomm bailing out is le bad.

... that the public is aware of.

If they had any Pat would shatter his voice chords on every ER call. Be serious.

 

[Hermitian]

If they had any Pat would shatter his voice chords on every ER call. Be serious.

No U.S. company talks publicly about black ops procurements.

 

[adroc_thurston]

No U.S. company talks publicly about black ops procurements.

The what? Anything government/military is miserable volumes. They need Actual Real Customers. Like Qualcomm. Or Nvidia. Or AMD. Or Marvell. Someone. Anyone with the ability to drive wafer (and hopefully packaging) volumes.
Again, be serious.

 

[Hermitian]

The what? Anything government/military is miserable volumes. They need Actual Real Customers. Like Qualcomm. Or Nvidia. Or AMD. Or Marvell. Someone. Anyone with the ability to drive wafer (and hopefully packaging) volumes.
Again, be serious.

Check the pricing of supercomputers.

 

[adroc_thurston]

Check the pricing of supercomputers.

They're built with commodity parts? NNSA systems like El Capitan and Crossroads are super boring off the shelf thingies.
Oh, and they're cheap.
El Cap is like 600m. Joke money by hyperscaler standards.
Be. Serious.

 

[Hermitian]

I see you are not familiar with the details and have a tendency to be flippant.

 

[adroc_thurston]

I see you are not familiar with the details and have a tendency to be flippant.

There are no details.
They do not have any anchor customers since 18A turned out to be very, well, mediocre.
They just have to execute on 18A for their own products now and maybe 14A will get some real deals.

 

[cebri1]

We got it. Intel has no costumers.

 

[jpiniero]

There are no details.
They do not have any anchor customers since 18A turned out to be very, well, mediocre.
They just have to execute on 18A for their own products now and maybe 14A will get some real deals.

Intel's not going to last that long.