Hardware

When asked whether 18A is more geared towards high-performance compute (HPC) applications than anything else, Chandrasekaran said this was "absolutely right."

"There are certain aspects of 18A that's extremely powerful for compute applications, especially the backside power. It's going to be very beneficial for compute applications."

But he added that all the "learning" from implementing that process node will go into its successor technology, 14A.

"So as 14A comes in, there will be a broader market that 14A will address, including compute and mobile and other applications and also how the PDKs (process design kits) are done so that it's not just for with Intel, but it's also focused on the broader ecosystem taking 14A and applying it to their designs," Chandrasekaran said.

Does this mean 18A will be relegated to high margin server CPU (and some high-end consumer desktop/mobile SKUs?)

"I think the big thing that's influencing margins next year is Lunar Lake. We have Lunar Lake, which is largely fabbed outside. There is a component of it that is fabbed inside. It's got memory in the package and we're just a pass through, couldn't get good deal on memory. So that suppresses the margins. And depending on how that product goes, it's kind of a headwind to us on the gross margin front," he explained.

Looking ahead, Zinsner said this will change. "Panther Lake is an 18A, or has an 18A component in it. So we start to see wafers come back. So we'll see this memory thing go away. We'll see more wafers going internal. We'll do better in terms of our cash cost per wafer. So that should be a nice tailwind for gross margins for us," he claimed.

(ignoring the memory aspect) Wasn't part of the point of decoupling design and manufacturing to make it not really matter 1) where the design team is doing manufacturing (pick the best one whether it's theirs or not) and 2) what designs the manufacturing team is producing? If they're so happy to get the manufacturing done in-house anyway, that seems contradictory to the overall strategy of splitting the two halves apart?