The US hopes to see TSMC holding a 20% stake in IFS (Intel Foundry Services) through a technology valuation or an actual cash investment.
On any case, if Broadcom would take Intel design, Broadcom needs to get x86 license and continue developing it.
IMO AMD knows this. Hope so.
Plus royalties for every x86 CPU sold. AMD had a huge part in advancing x86 technology so no reason they shouldn't benefit from it.And at the end, AMD should ask Broadcom to share with AMD its networking IP in return for being able to use x86 license.
Plus royalties for every x86 CPU sold. AMD had a huge part in advancing x86 technology so no reason they shouldn't benefit from it.
10+ years too late for that. Who in their sober minds would licence x86 now for a brand new future processor?Heh. Intel could always become like ARM.
Samsung?Who in their sober minds would licence x86 now for a brand new future processor?
Already working with AMD. So no... the only option? China...Samsung?
These are the figures which AMD presented for Q4/2024:AMD is already selling more server CPUs and desktop is coming. Laptop ?? Not sure.
Its somewhere in this forum. I could have also got the wording wrong.These are the figures which AMD presented for Q4/2024:
View attachment 117344
(via computerbase.de)
Now I don't know how the remaining non-AMD market share splits into Intel and non-Intel percentages.
These are the figures which AMD presented for Q4/2024:
View attachment 117344
(via computerbase.de)
Now I don't know how the remaining non-AMD market share splits into Intel and non-Intel percentages.
[edit: as Markfw points out below, the table has been posted before: AMD 2024-Q4 Earnings, page 3]
I hope that changes. Intel has had crap for DC CPUs for years now.AMD already sells more datacenter CPUs to US hyperscalers, but Intel sells more to "enterprise"
It was easy for Pat coz he was sleeping ON the jobOn top of all of these requirements you have to be the type of person that can, at the same time, both make hard decisions/fire people and sleep at night, while still being compassionate.
“The U.S. government wants the U.S. to have large capacity in the U.S. at under 2 nanometers, and this is appropriate. The key is the level of commitment that TSMC will make to capacity in the U.S. TSMC has no interest in the wafer-fab facilities of Intel. We talk with both companies on an almost real-time basis.”
There is no chance of an agreement, Hutcheson said.“There is no reason for TSMC to help Intel, other than the uncertainty around President Trump’s ask. Intel can manage its own fabs, and 18A is coming along nicely. What Intel needs is to fill its fabs.”
“It’s only good in the interest of Intel’s self-inflicted financial situation. It hurts America’s national-security interests. It is not in TSMC’s interests. Why save your drowning chief competitor when you’ll have a monopoly once they’re gone?”
“TSMC is in the most delicate of geopolitical positions and has to answer the phone when Washington calls, however bad the request could be.”
“Who decides which customers use which foundry services? This is not a simple question, as it would involve a lot more collaboration between Intel, TSMC and leading design houses. Trump administration officials have not had enough time to look at the issue in sufficient depth, and this effort is very preliminary.”
The administration will urge TSMC to increase its currently planned $65 billion investment in two fabs in Arizona to at least $200 billion and five fabs, Huang said in a post on LinkedIn. That would include a push for TSMC to move its CoWoS advanced-packaging technology to the U.S., he added. In Taiwan, TSMC uses CoWoS to make AI chips for Nvidia and a handful of other chip designers.
I had high hopes for Pat. He may have gotten a raw deal in that he wasn't provided enough time for his vision to come to fruition. Or he may not have been well rounded enough in all of the traits I mentioned above, perhaps too heavy on the tech and not enough on economics and people. Who knows? But it's not for lack of intelligence these people don't come through. They are all quite smart, it's having the entire skill set in one individual, the passion, and the work ethic.It was easy for Pat coz he was sleeping ON the job
That's orthogonal to the point. All the platform advancements mentioned as having been Intel's handiwork wouldn't be coming from Apple or Qualcomm, either (well maybe Apple, but they'd keep everything proprietary for as long as possible). All the implementation support mentioned as an advantage when working with Intel would definitely not come from Apple (vertically-integrated) or Qualcomm.That would leave Apple and Qualcomm as the biggest winners.
One of the problems would be capacity.That's orthogonal to the point. All the platform advancements mentioned as having been Intel's handiwork wouldn't be coming from Apple or Qualcomm, either (well maybe Apple, but they'd keep everything proprietary for as long as possible). All the implementation support mentioned as an advantage when working with Intel would definitely not come from Apple (vertically-integrated) or Qualcomm.
All that aside, if AMD wants to continue keeping x86 alive, they're more than capable of doing it by themselves. There are some other things people might miss if/when Intel is finished as an IDM, but x86 won't be one of those things.
Why does this even matter? This is a non-story. If the cross licensing agreement automatically cancels, AMD can't make x86 CPUs either. It would be in the best interest between AMD and Intel's acquirer to hash out a new cross licensing agreement.According to Retired Engineer (or ChatGPT), revocation of cross license is automatic on change of control. AMD would not have to do anything.
Why does this even matter? This is a non-story. If the cross licensing agreement automatically cancels, AMD can't make x86 CPUs either. It would be in the best interest between AMD and Intel's acquirer to hash out a new cross licensing agreement.
Well, was this mutual (in case of acquisition) or only targeting AMD ?
One of the problems would be capacity.
AMD design capabilities are not in question (AFAIK) but if Intel in one year (let's say) doesn't ship x86 anymore, how would AMD be able to compensate ?
They won't, not for quite a time, ordering wafers and packaging isn't like fast food as we all know.
And then less competition would be bad too.
And this analysis comes from ChatGPT?Why does this even matter? This is a non-story. If the cross licensing agreement automatically cancels, AMD can't make x86 CPUs either. It would be in the best interest between AMD and Intel's acquirer to hash out a new cross licensing agreement.