>. The only way to justify the research costs and fab costs is to amortize it across more chips
or make more money per chip, which is what intel does, since lets say compared to an AMD chip TSMC manufacturers, TSMC takes a cut of the chip for manufacturing and then AMD takes the rest, while intel collects both portions.
In return it also has two sets of R&D to support and two sets of risks - architecture and manufacturing. If it falls behind on either of these it starts to lose.
TSMC for example can solely focus on manufacturing assured that it will fill it’s fabs if it keeps pace.
Maybe Intel made super profits when x86 was the only game in town but that’s not the case any more.
>If it falls behind on either of these it starts to lose.
If TMSC falls behind in one of these they lose, and they don't have the other. Is that an advantage as you seem to put it? If they make a wrong choice like intel did for 10nm they're going to be immediately a non-entity with no 'other' business. Having two sets of money make businesses puts intel at a big advantage in terms of financing and owning their own platforms.
If TSMC falls behind a node all of their orders will disappear to whoever has a more advanced node. They don't have another business. Instead of two risks, they have one risk thats identical to intel's, and their entire business depends on it. That's a lot less anti fragile.
Intel has two sets of risks and in exchange on many many fewer chips they basically made the same amount of money last year, when they were behind on CPUs at almost every metric. That's resilient. People talking about the fall of intel are talking about something that intel is actively maneuvering ahead of. TSMC has no chip design risks much lower per chip profits in exchange.
Vertical integration is great if it generates synergies. It’s really bad if the tie into the internal customer hinders the development of each part of the business.
Intel is not remotely robust as it’s almost completely dependent on x86 and needs to catch up with TSMC. It lost smartphones in part because of x86. Now it’s fallen behind AMD because of manufacturing weaknesses. Hence a P/E ratio of 9 vs c20 for TSMC.
TSMC on the other hand has a huge variety of customers at a wide variety of nodes.
Intel never had smartphones to lose. Inetl can generate all the same advantages as amd by simply buying Chips from TSMC if they want (and they already do for some chipsets) so there is no operational disadvantage. Intel is already mostly caught up to amd and will be making significantly more per chip than amd very soon.
TSMC is competing for something Intel doesn't want to sell. Intel even when it was in the lead wasn't tabbing it's newest process for 3rd parties. You're declaring TSMC Victor in a game Intel never played. And in a handful of years if Intel gains back process advantages you will likely still declare them the loser for not playing the cutting edge fab for other companies game they don't want to play.
TSMC is not playing the same game Intel is, and in 2021 when by all accounts Intel was behind TSMC and and, they still managed to make similar profit to TSMC and laugh at AMD's inability to buy enough chips to make anything close to competition for either Intel or Nvidia.
Now they're also getting into graphics cards and have largely caught uP with amd designs. Their future is bright.
Intel has consistently tried to build a top tier GPU and failed year after year. Expecting them to suddenly break away from their history is extremely optimistic.
They're just starting to get into the business. It will improve. They don't have to have the best cards, they just have to compete in some segments from the start. It's all upwards from here.
That's the thing about experience - you keep accumulating it.
or make more money per chip, which is what intel does, since lets say compared to an AMD chip TSMC manufacturers, TSMC takes a cut of the chip for manufacturing and then AMD takes the rest, while intel collects both portions.
reply