Funny you mention this. In my brief stint there, I saw a fresh college graduate get promoted to a lead for a Scanning Unit simply because the current lead was retiring (I was actually offered that position, but I was leaving and turned it down). They were trained in less than a month by shadowing the lead on-the-verge-of-retirement. The engineer who got promoted was at Intel less than a year, and had no prior internship experience (they were hired in 2021 when chips were in desperate need of talent. You might recall the chip shortage that affected cars etc.)

> A few years ago a CEO of Intel (not Gelsinger) said something like "our CPU architecture is so well-known every college student can work on it". A friend working at Intel at that time translated it to me: "we will hire cheap students to work on chips and we will let the expensive engineers leave"

Reminds me of the Boeing managers saying that they didn't need senior engineers because its products were mature..

A few blown doors and deadly crashes later, that didn't age well..

https://archive.is/H9eh9#selection-3935.0-3935.163

I know some people who understand x86 [0] very well. Most of them do not work at Intel. Those that do tend to be on the OSS side and don’t really have any sway in the parts of Intel that design new hardware.

And this is a problem! Most of Intel’s recent major architectural changes over the last decade or so have been flops. [1] Quite a few have been reverted, often even after they ship. I presume that Intel does not actually have many people who are really qualified to work on the architecture.

[0] I’m talking about how it interacts with an OS and how you put it together into a working system. Understand stuff like SIMD is a different thing.

[1] AVX and its successors are notable exceptions, but they still have issues, and Intel did not really rock the early implementations.