[Throwawayaerlei]

A bit on how this played out: what they call their first 10nm node which in its third iteration has been named Intel 7 was more aggressive than TSMC's first 7 nm node, and failed for many years to economically produce chips. Both use 193 nm UV lithography, and then TSMC made a more aggressive than Intel's node using some EUV from ASML, and both TSMC nodes worked.

Now TSMC is two major nodes ahead with 3 nm risk production credibly scheduled for this year and mass production next year, with Intel said to be buying a lot of that. What was Intel's still delayed 7 nm node, the first to use EUV, is now named Intel 4. I've not looked at it closely, but it looks like something equivalent to TSMC 3 nm nodes is scheduled as Intel 20A for angstrom and 18A, sometime in 2024 and 2025.

I have yet to see anything that convinces me Intel will regain its ability to make state of the art logic chips, which for many generations was one of their most important advantages, allowing them to beat "smarter" CPU designs with their own CPUs being manufactured 1-2 nodes ahead of everyone else. It will be interesting if some day the true story of how this happened is revealed.

[DSingularity]

Doesn’t this mean the Intel is done? Intel has never designed chips with innovation that matches the quality of AMD, Apple, and Nvidia. I just don’t see how they survive as a fabless.

[Throwawayaerlei]

Doesn’t this mean the Intel is done? Intel has never designed chips with innovation that matches the quality of AMD, Apple, and Nvidia.

That's frankly not true, for example see them along with a few others not including the above companies reviving the out-of-order technology IBM first developed for its System 360 supercomputers, for Intel first in the Pentium Pro.

Culturally I doubt they can shift to fabless although we'll be able to tell if they're really trying by seeing mass layoffs of managers, and I've very concerned they didn't recruit a fab guy to be their CEO although I wonder if no one qualified was willing to take the position. Another argument against that is their placeholder CEO who'd previously been their CFO actually had a clue, I suppose wasn't part of one of the factions involved in the fab debacles, decided to go fabless, and his strategy was junked by the new CEO.

In general I don't see a path to victory for them given my guess that they won't be able to regain their ability to move to new nodes that can make economical chips. Although I haven't investigated their third iteration of their 10 nm node, now Intel 7, need to spend more time reading Semiwiki etc. and also see if there's any sign they're going to get their first EUV node, Intel 4, working in any practical time frame, it's already delayed.

On the other hand, AMD has done it again, spent a huge amount of their capital on this time an FPGA company; at best that means they believe they don't have anything better to devote resources to, which is bad news for their AMD64 ecosystem. If you've followed their history going back decades, it was said a while ago over the long term they've never made money for their stockholders, and they've repeatedly gained serious leads only to blow them, raise your hand if you once used one of the 100MHz 486DX4s. Just like Intel's cultural problems, I suspect this aspect of AMD's culture will give Intel breathing room sooner or later, as it did when K8 got long in the tooth.

Apple isn't even in the running unless they decide to start selling their ARM SoCs to third parties. Who gives a damn about those if you're not in the Apple ecosystem, which has it's own problems. Nvidia if allowed to buy ARM will I expect destroy that ecosystem, certainly make it uncompetitive, including eliminating competition they would be facing thus likely making them complacent.

[DSingularity]

So what is your final say? You say they have a chance at executing an IBM scale pivot, but a low one? You also say that AMD historic inability to generate profits will give them some breathing room?

So, intel will recover because it will gain breathing room and it will benefit from a pivot?

It seems like a stretch.

For one it seems like shifting a company used to an accommodating internal node users to one that has to retool and shift to TSMC/Samsung is a major ask. For two it seems like major profits are around the corner for AMD. This is a company that has historically had <1 % of the markets intel participate in. Those numbers have grown by factors and there is no sign they will stop.

Would I count Intel out? Maybe not. Maybe the US gov will provide a lifeline. But there is not a doubt in my mind that Intel could stumble again and loss the manufacturing advantage permanently. That seems like it would be a mistake that would lead to another 5 years of market share loss.

[Throwawayaerlei]

I certainly have no "final say," I'm just pointing out some of the forces including internal cultural ones that I believe may have effects on the outcomes.

IBM has nothing to do with the business or technical questions, I'm referring to the https://en.wikipedia.org/wiki/Tomasulo_algorithm which was first used in the late 1960s in the floating point unit of the IBM System/360 91 and much later used for the Pentium Pro.

AMD's historical inability to generate profits pertains to its historical inability to succeed well for very long, it's an independent measure of that phenomena. So no matter how well AMD is doing right now, it's in their DNA to fail any time now; doesn't mean it's going to happen, but I'll repeat again that paying a huge sum for an FPGA company is a very bad sign, much worse then their purchase of ATI which did not end up helping their CPU efforts, if anything distracted from the latter. Another sign AMD "will stop" is the reported generally lower quality ecosystem, but I have no good reading on that.

Your comment about a "major ask" is just repeating why I previously said, "Culturally I doubt they can shift to fabless although we'll be able to tell if they're really trying by seeing mass layoffs of managers...."

How much of AMD's "major profits around the corner" will be put into the FPGA unit? It was an all stock deal so they at least won't be having to pay down debt on it, but to restate my main point about it, it's a vote of no confidence in their CPU business, doing it comes at tremendous opportunity costs.

Never heard that money was the core problem for Intel's failure to move to new nodes, so the US government doesn't have the right sort of lifeline to provide. Intel is currently stumbling and I personally guess that won't stop, but that's based on very thin data and what I see the new CEO trying to do.

Or not, if the denials about the rumor about them buying Global Foundries directly from its owner Mubadala Investment Company are correct. If managed well, which is never the way to bet in tech company acquisitions, that would give them invaluable expertise in the business of being a foundry.