Now former Intel CEO Pat Gelsinger confirmed before the launch of Arrow Lake-S that they will not end up using Intel’s 20A node in this new generation of CPUs, saying that They preferred to leave it so that we can devote all available resources to the development and improvement of the Intel 18A node, a node that will play a critical role in the company.

Not much explanation was offered at the time, but Intel said the work done with the 20A node contributed to the development of node 18Aand this allowed us to discover things that were very important for the betterment of the other. In short, this means that all the funds invested in the canceled node were not, at least in theory, a waste.

It should also be noted that the 20A node was used to implement many advances such as transistors RibbonFet Gate-All-Around (GAA)which will replace the FinFET transistors used since 2011. Intel has been producing Arrow Lake processor wafers with a 20A node, so this has become fully functional, although we have no details of its success on the wafer.

The chip giant bet everything on the Intel 18A node, and according to South Korean sources, this move could be a mistake, as at the moment the success rate on the wafer that has this node is so low that it would not even reach 10%. That is, of all the chips coming off the wafer below that node less than 10% of the work.

I’ll give you an example, imagine you can get 50 chips made under an Intel 18A node on one wafer. A 10% success rate means that out of those 50 chips, only 5 would work and the rest would be discarded. The production cost of a wafer is fixed and its profitability depends on the total number of functional chips that can be obtained from it, which means that the loss of success translates into very high costs.

With a success rate of less than 10% an Intel 18A node would not be viableand this is a problem, because as I said before, this node must become one of the great pillars of the chip giant and will be the foundation that will give life to other processors for servers and data centers from Intel, Clearwater Forest. family.

Intel still he has time to polish things and make the necessary adjustments to improve success on that node’s wafer, but it’s very likely that you’ll have to make drastic changes to achieve that, and that would complicate things a lot. We’ll see what happens.