OperandiThis is technology forum not a financial forum. It terms of technology AMD is giving Intel a pretty big beat down.

He said whooping, they are not. They might have a better CPU atm, but that will not last forever. In terms of financial, which is relevant, they are no where near whooping.

john_With GF stopping any further development in 7nm, I don't think it's affecting AMD.

The Ryzen IO dies are still made by GloFo

Seems like we are on the cusp of a new era of anti-competitiveness, where the predatory business mind takes over when the engineering execution leaves much to be desired.

I think this is an interesting move to secure patents, access interesting customers/new IP, have more foundries all over the world, acquire various old and new-ish lithography and processes for a variety of products (perhaps even foveros?), etc. TSMC could have pulled a similar stunt; it would be entirely befitting of their portfolio, too.

But best of all, I love the full circle irony. It would be poetic if the acquisition sank them.

R-T-BThe Ryzen IO dies are still made by GloFo

And probably some Polaris cards, some Ryzen 2000 chips, older EPYC chips, APUs, embedded Ryzen CPUs and who knows what else. But AMD probably already signed a deal with GF for the next one or two years that Intel probably can not cancel without paying a good amount of money. And even if they want to cancel that deal, until they get GF and takeover those factories, I guess some time will pass.
The thing is that AMD does not need GF's 12nm for much longer. With TSMC and Samsung going to 3nm, 7nm process probably will become much cheaper in the future, making it a viable alternative to GF's 12nm, even from a financial perspective. AMD is building Vega, RDNA, RDNA 2, Zen 2, Zen 3 and I/O(for mobile Ryzen chips) already at 7nm at TSMC. So, they don't really need to start redesigning stuff to 7nm. What AMD could have done all this time, but they didn't was to build APUs and low core count CPUs at 12nm GF. I mean, why not build 3400G and 3100, 3300 at 12nm and free 7nm capacity? Why not build mid/low end RDNA cards at 12nm GF to fill the mid/low end market? They probably had their reasons, but it proves that they don't really have much use of GF's 12nm.

UpgrayeddIs that why a backported architecture on a larger process is still competing with the cutting edge processes?

He is a shameless AMD fanboy and cannot think straight when it comes to these things, but I have a feeling he is going to be very upset when Alder Lake lands...

stimpy88Aren't GF even further behind in process tech than Intel?

A strange way to spend 30Bil.

Fab capacity and smaller nodes are two entirely different metrics and both are in high demand. Workforce is what they are buying, production lines fully staffed and built. Thats a major leap with 'direct' time to market, as opposed to say, TSMC setting up new plants in EU or US.

They will try to hinder AMD production. A dirty strategy.

All AMD production worth mentioning is at TSMC (where it has to be as they have the most advanced node(s) that team red needs to balance out its designs which are still trailing behind), so that is not a valid argument whatsoever.

mclaren85They will try to hinder AMD production. A dirty strategy.

AMD is using TSMC and is on smaller nodes than GF can use.

The hindrance is probably a competition on price using inferior chips, much like the game AMD has tried to play not too long ago. A loser's strategy that can work to stay afloat until the next design win.

HenrySomeoneHe is a shameless AMD fanboy and cannot think straight when it comes to these things, but I have a feeling he is going to be very upset when Alder Lake lands...

By the time Alder lake comes out Zen 3 will have been on the market for over a year. If Alder Lake isn't better in at least some metric its going be mega embarrassing. Personally I think Alder Lake will have a good enough of an IPC lead and clock speed advantage that it should win on lightly threaded loads and probably most moderately threaded loads as well. It will probably blow through a shit load of power to do it though.

UpgrayeddIs that why a backported architecture on a larger process is still competing with the cutting edge processes?

There seems to be some misunderstanding about about what 7, 10 and sub nm nodes are net you. 7/10nm vs. 14nm gets your more density and lower power consumption in terms of CPU or GPU. Aside from that both Intel and AMD could have stayed on 14nm and kept on producing 8 core chips and advancing their designs. AMD would still be ahead with Zen 3, there just wouldn't be 16 core chips on mainstream platforms.

The other aspect is in terms of production it gets you more dies per wafer. So Intel gets less product per wafer and is pushing the core limits making huge dies on lager node process, so just overall less competitive from cost to manufacture sense compared with what AMD / TSMC are doing.

OperandiBy the time Alder lake comes out Zen 3 will have been on the market for over a year. If Alder Lake isn't better in at least some metric its going be mega embarrassing. Personally I think Alder Lake will have a good enough of an IPC lead and clock speed advantage that it should win on lightly threaded loads and probably most moderately threaded loads as well. It will probably blow through a shit load of power to do it though.

There seems to be some misunderstanding about about what 7, 10 and sub nm nodes are net you. 7/10nm vs. 14nm gets your more density and lower power consumption in terms of CPU or GPU. Aside from that both Intel and AMD could have stayed on 14nm and kept on producing 8 core chips and advancing their designs. AMD would still be ahead with Zen 3, there just wouldn't be 16 core chips on mainstream platforms.

The other aspect is in terms of production it gets you more dies per wafer. So Intel gets less product per wafer and is pushing the core limits making huge dies on lager node process, so just overall less competitive from cost to manufacture sense compared with what AMD / TSMC are doing.

Lol even with the zen hype amd will never break 50% market share vs intel..

RichardsLol even with the zen hype amd will never break 50% market share vs intel..

LOLOLOL, I give zero fucks what the majority of the market does, I buy whoever has the best shit out at the moment.

Just remember how fickle PC gamers are, they will buy the best whether it is AMD or Intel. So if they do release a AMD beating chip, it is only going to be bad for AMD

this is big news.
but isn't this creating a monopoly on US ?

No way regulators allow this monopoly.

RedwoodzNo way regulators allow this monopoly.

TSMC dominates the foundry business - manufacturing chips for its customers. Almost all if not all of Intel's current foundry production is internal - it manufactures chips for the other business units inside the company. Same applies to Samsung which does some foundry business but lion's share of its production goes to its own other business units.

Most analytics on foundry business go by revenue so it is probably a little more even in terms of output in wafers but top players have considerably larger margins on state of the art nodes (and the accompanying R&D expenses). Foundry business shares look roughly like this:
www.trendforce.com/presscenter/news/20210531-10809.html


Edit:
By wafer capacity Intel is left out of top5 with 884K wafers/month. This is total installed capacity, does not exactly show not actual production - which should be close-ish to max - or foundry business share - see the note above about Intel and Samsung.
epsnews.com/2021/02/10/5-fabs-own-54-of-global-semiconductor-capacity/

Didn't AMD divest GlobalFoundries and decided to go fabless, probably to overcome their former struggles?
This is huge if a CPU designer improves its manufacturing capabilities, we might not see a shortage ever again but competition instead, TSMC is incompetent and can't handle anything alone.

ZoneDymooh no, that is not good, I hope regulators dont let that happen.

Amazon has gone and bought everything it wanted and still does, where are your regulators? :)