How important is gloflo to AMD and Apple and others making their own/ARM designs etc? Would this even be allowed under anti-trust/competition regulations?

ik9000
How important is gloflo to AMD and Apple and others making their own/ARM designs etc?

Not very, TSMC and Samsung are bigger players for them.

Would this even be allowed under anti-trust/competition regulations?

I don't see why not. It used to be solely under the AMD umbrella in the past.

kalniel
Not very, TSMC and Samsung are bigger players for them.

I don't see why not. It used to be solely under the AMD umbrella in the past.

But didn't Intel own their own fabs too? Presumably AMD were also the minority player in that scenario? I don't see how that's the same.

IIRC GlobalFoundaries was originally AMD's internal fabrication unit before it spun it off to raise capital a little over 10 years ago, so a little ironic to see Intel sniffing around.

On the other hand, GF decided that heading to 7nm would be too expensive when there was still plenty of demand for chips made at 14nm (again IIRC - I may be muddled). As such, GF basically left the race a few years back, in terms of being able to make modern CPUs. In recent years the number of top tier fab options has dropped significantly from about 8 to about 3. Again, I haven't checked those numbers. As others have said, AMD moved to TMSC for Zen and RDNA product lines.

Whilst Intel probably do have a use for more 14nm manufacturing, it'd take a lot of investment to bring GF down to 7nm (or lower). Unless they have a very specific reason/requirement to purchase GF, investing in new facilities or upgrading existing ones would almost certainly be more cost effective in the long haul. Of course, cost effectiveness isn't the be all and end all, though!

Irien
IIRC GlobalFoundaries was originally AMD's internal fabrication unit before it spun it off to raise capital a little over 10 years ago, so a little ironic to see Intel sniffing around.

Sort of. AMD did get rid of their fabs to GF, but I'm pretty sure it cost them money to do so with all sorts of payment schedules and wafer usage guarantees in place.

But GF have slurped up other fabs along the way as well. Stuff like IBM, so it isn't just the old AMD fabs any more. A lot of wafers have passed under the stepper since the split.

This does makes some sense for Intel. They can't make much more money on CPUs, they are trying to move into fab services. Buying up an existing fab business would give them a new revenue stream, and instant flexibility in offering things like SOI and old nodes to customers that won't want Intel 14nm+++ parts even if Intel had lots of spare capacity (which they plainly don't). It also plays to Intel's main strength as a manufacturer coming from a bulk dram background. Let's face it, Intel have generally made the best x86 CPUs out there, but compared to other architectures their designs have always been a bit rubbish but bolstered by their manufacturing ability.

At this point, I think AMD are only really making the Ryzen memory controller die on 12nm and I imagine that is an old enough bit of silicon that they have migration plans in place already. I think there are plenty of other ways to say this is bad for the global economy though.

Irien
IIRC GlobalFoundaries was originally AMD's internal fabrication unit before it spun it off to raise capital a little over 10 years ago, so a little ironic to see Intel sniffing around.

On the other hand, GF decided that heading to 7nm would be too expensive when there was still plenty of demand for chips made at 14nm (again IIRC - I may be muddled). As such, GF basically left the race a few years back, in terms of being able to make modern CPUs. In recent years the number of top tier fab options has dropped significantly from about 8 to about 3. Again, I haven't checked those numbers. As others have said, AMD moved to TMSC for Zen and RDNA product lines.

Whilst Intel probably do have a use for more 14nm manufacturing, it'd take a lot of investment to bring GF down to 7nm (or lower). Unless they have a very specific reason/requirement to purchase GF, investing in new facilities or upgrading existing ones would almost certainly be more cost effective in the long haul. Of course, cost effectiveness isn't the be all and end all, though!

It could be more about getting the contracts/customers? Its also gives them a pool of trained engineers?

cheesemp
It could be more about getting the contracts/customers? Its also gives them a pool of trained engineers?

Probably of more value than that would be any patents owned by GF that could be used to squeeze easy revenue from TSMC or Samsung.

It feels the only reason Intel might gobble up GloFlo is so they can get the expertise of selling wafer/fab space to potential customers which Intel has practically never done at scale.

But I also expect this to have very serious observations by competition authorities in various nations as it will be dropping the big four silicon manufacturers down to three. Although GloFlo may no longer be fighting on the bleeding edge, they are still a reasonably substantial in their manufacturing silicon globally.

Intel seems to just be looking at hoovering anything up to realise their vision.

GF does SOI nodes,which are useful for specialist commercial,space and military applications. ATM,GF is foreign owned and getting it back into US ownership again might have some NatSec angles which the US government might want to facilitate. Remember,IBM gave its expertise for free to GF,and why they are involved in legal wranglings so its not only GF expertise,but also stuff IBM has researched too. So if Intel knows it can potentially get companies such as IBM onboard if it can offer specialist nodes on top of what it already offers.

CAT-THE-FIFTH
GF does SOI nodes,which are useful for specialist commercial,space and military applications. ATM,GF is foreign owned and getting it back into US ownership again might have some NatSec angles which the US government might want to facilitate. Remember,IBM gave its expertise for free to GF,and why they are involved in legal wranglings so its not only GF expertise,but also stuff IBM has researched too. So if Intel knows it can potentially get companies such as IBM onboard if it can offer specialist nodes on top of what it already offers.

That's a very good point, it will increase their portfolio in those areas. However, it also could depend on what happens between GloFlo and IBM on those areas being litigated on.

However, as Intel has had very little dealings as an IDM, they have no infrastructure, process or any systems in place to facilitate selling outside of internally and setting those up could take a substantial amount of time. But then again, purchasing a company as large as GloFlo it could take just as long to transplant that function into their own business portfolio.