Intel really needs to be studied for how it's spiralled into a possibly terminal state of dysfunction.
EPIC was Intel's first big folly that I remember and it almost killed the company. Years late, too expensive, motivated by Intel's desires rather than customer needs, software would've required significant rewrites at every level. The Gigahertz race (culminating in the Pentium 4) was near-fatal to AMD but then near-fatal to Intel and they only recovered because somebody went back to the Pentium 3 and bolted on AMD's x86_64 extensions. Still, for half a decade Athlon/Opteron were a huge problem for Intel..
But the next near death experience came in the 2010s with the years-delayed move from 14nm to 10nm processes. This came kind of a joke in the industry and it's honestly still not clear to me what went wrong. Intel had previously migrated to smaller and smaller processes like clockwork.
Many (including myself) thought Pat's appointment was a good sign for Intel, being an engineer. From the outside it seemed like Intel was organizationally broken, having descended into fiefdoms that didn't communicate let alone cooperate.
So this conversation is largely about what Pat's doing now rather than what happened at Intel. He's previously pointed the finger at Wall Street "short-termism" [1], for example.
It's interesting that they talk about the relationship with the current administration being positive and receiving funding from the CHIPS Act. The CHIPS Act was passed in 2022.
But if we're looking ahead here, I'm a little disappointed China was only mentioned once at that was just with building solar capacity. I'm sorry but if you're in the chip space and not thinking about what China is doing, I don't know what you're doing.
China has been restricted by the US from buying the best chips and buying the latest lithography technology. So they're making their own and it's happening way faster than I think many of these industry types are telling you it will happen. A few years from now when China produces EUV chips, I guarantee you'll read news articles about how this came out of nowhere and nobody expected it.
[1]: https://wccftech.com/intel-ex-ceo-blames-wall-street-for-ame...
> But the next near death experience came in the 2010s with the years-delayed move from 14nm to 10nm processes. This came kind of a joke in the industry and it's honestly still not clear to me what went wrong. Intel had previously migrated to smaller and smaller processes like clockwork.
I'd love to see a post-mortem. Intel's 14nm came out with Broadwell in 2014, and it was their desktop/server node for seven years, until Alder Lake in 2021. But they got stuck on the micro-arch as well. They shipped what was essentially the same Skylake core for five years from 2015 to 2020.