If this sale were to actually happen AMD will laugh all the way to the bank and I’d bet quite a few would object.
I agree that this is trolling pure and simple and shows how shallow, superficial the business press is who doesn’t understand the technical nuances and or the sense to ask analysts who do.
Very good point! I forgot to mention that Intel's AMD64 license is non-transferable. AMD has historically been friends with Qualcomm and a deal would plausible.
Agree the mainstream business has completely missed this. Crazy how WSJ is covering this story as if it is legit news.
The cross-licensing agreement with AMD would be null in the event of a change of control. But the counterpoint to this is Intel licenses the right to make x86 CPUs to AMD in exchange for the right to make x86_64 CPUs.
Technically, the license will terminate. But in practice, AMD would be compelled to re-establish the cross-licensing agreement with new owners.
But the interesting question is whether AMD could jettison 32 bit x86 and just make AMD64 instruction set only CPUs or is that considered a derivative work on top of x86? One could easily skip all 32 bit compatibility these days as Windows doesn't even ship a 32 bit version anymore. Just binary translate to AMD64 at runtime and the performance should be good enough for older software with the latest chips.
Regardless of the legal nuances, I would say that the 64 bit extension of x86 would be the more valuable part.
You are doing God’s work! Really enjoy your informative and timely pieces!
How do you see Broadcom acquiring Intel? (In Bloomberg’s updated article, Broadcom is said to have considered buying Intel, even though imo there’s prolly a very low chance that such deal goes through)
I could see AVGO buying Intel Design. NEX would be mostly complimentary to the existing Broadcom networking business. x86/CPU could be milked well by Hock Tan. Problem is, Intel Products market position is so weak that Hock would not be able to raise prices, only cut OpEx very deep.
I 100% agree with you on the first point. Ampere Computing would add datacenter software expertise, datacenter QCL experience, and a portfolio of high-performance memory and PCIe chiplets. Swap out the Ampere cores for Nuvia ones and you have a real HPC contender.
For driver/compiler I still think they can just hire/poach Intel engineers for much cheaper than any formal deal.
There are two types of M&A. The ones that raise substantial antitrust concerns and those that don't. The ones that raise antitrust concerns, BUT CAN BE APPROVED, are generally good M&A. But why on earth would you do large scale M&A with little antitrust concerns??? And when I looked at the news I thought "this could be approved" and laughed.
On top of everything you say, Intel cannot sell the x86 license to Qualcomm -
https://x.com/outspokengeek/status/1830019248519070131?s=46&t=W4ElQT-wX8qfoIMltPWnKg
If this sale were to actually happen AMD will laugh all the way to the bank and I’d bet quite a few would object.
I agree that this is trolling pure and simple and shows how shallow, superficial the business press is who doesn’t understand the technical nuances and or the sense to ask analysts who do.
Very good point! I forgot to mention that Intel's AMD64 license is non-transferable. AMD has historically been friends with Qualcomm and a deal would plausible.
Agree the mainstream business has completely missed this. Crazy how WSJ is covering this story as if it is legit news.
The cross-licensing agreement with AMD would be null in the event of a change of control. But the counterpoint to this is Intel licenses the right to make x86 CPUs to AMD in exchange for the right to make x86_64 CPUs.
Technically, the license will terminate. But in practice, AMD would be compelled to re-establish the cross-licensing agreement with new owners.
Possible. I'm not a lawyer.
But the interesting question is whether AMD could jettison 32 bit x86 and just make AMD64 instruction set only CPUs or is that considered a derivative work on top of x86? One could easily skip all 32 bit compatibility these days as Windows doesn't even ship a 32 bit version anymore. Just binary translate to AMD64 at runtime and the performance should be good enough for older software with the latest chips.
Regardless of the legal nuances, I would say that the 64 bit extension of x86 would be the more valuable part.
You are doing God’s work! Really enjoy your informative and timely pieces!
How do you see Broadcom acquiring Intel? (In Bloomberg’s updated article, Broadcom is said to have considered buying Intel, even though imo there’s prolly a very low chance that such deal goes through)
https://www.bloomberg.com/news/articles/2024-09-21/chipmaker-qualcomm-is-said-to-explore-friendly-takeover-of-intel
I could see AVGO buying Intel Design. NEX would be mostly complimentary to the existing Broadcom networking business. x86/CPU could be milked well by Hock Tan. Problem is, Intel Products market position is so weak that Hock would not be able to raise prices, only cut OpEx very deep.
Ampere makes more sense for Qualcomm to acquire than Intel.
This announcement is so strange, it really does feel like a meme or miscommunication.
I am thinking Qualcomm might want to acquire Intel’s software focused teams like driver development or compiler support.
I 100% agree with you on the first point. Ampere Computing would add datacenter software expertise, datacenter QCL experience, and a portfolio of high-performance memory and PCIe chiplets. Swap out the Ampere cores for Nuvia ones and you have a real HPC contender.
For driver/compiler I still think they can just hire/poach Intel engineers for much cheaper than any formal deal.
Thank you for your work, sir.
There are two types of M&A. The ones that raise substantial antitrust concerns and those that don't. The ones that raise antitrust concerns, BUT CAN BE APPROVED, are generally good M&A. But why on earth would you do large scale M&A with little antitrust concerns??? And when I looked at the news I thought "this could be approved" and laughed.