Well, Intel had a strategy and a CEO. The CEO was forced out by the board, but we don't know why. All we've heard is the board wasn't happy with the direction Intel was taking. How long have they been unhappy? How much infighting has there been on the board? What happens to Intel's CHIPS award? What happens to the foundries whose construction was already underway? Is Intel really in a complete disarray?
I think there's some deep seated cultural issues in Intel and Gelsinger wasn't the person to fix them. I have friends who work there and it sounds like a different planet to the semiconductor company I work at. Intel does have the bonus of being 'too big to fail' from a US security point of view so I imagine those fabs will get finished one way or another.
Also, it's very laid back, but Bryan Cantrill's podcast touches on some of these issues. Interesting to see the perspective of an Intel customer [1].
The board wants to split the company, the CEO didn't. That's what the fight was about. Should Intel become a design company? And have TSMC do everything? IMHO, nope. The rot in Intel happened a long time ago, the smart people left a long time ago, and the culture changed for the worse. They would have to spend ridiculous amount of money to lure smart folks from Nvidia & AMD, etc. Then the difficult battle of building a winning culture. I was long intel for a while. I unloaded all my stocks this year. I don't believe anymore, and I'm not even convinced with the CEO change.
they and TSMC don't compete with ASML.
If you have the time, read "Chip wars" that will definitely convince you that USA needs a local chip maker and a Fab if the wish to retain their superpower position. Prior to reading the book, I'll agree and say to hell with the fabs. Just design and let TSMC make it. After reading the book, USA needs a miracle.
Competing with TSMC using ASML’s kit means beating them on input costs, operations and/or distribution. That’s doable but tougher and more ephemeral than having a technological moat. (It’s Herculean given TSMC’s scale.)
If someone wants to take on TSMC, they should be looking at leapfrogging or sidestepping ASML in a differentiating way. Not out of the gate. But in a PA Semi sort of way.
ASML has no competition, you will have an easier time replacing TSMC before doing ASML. Frankly, I think they should do a hybrid approach, have TSMC do some stuff and keep refining their own Fab till they figure it out.
I think you may have overlooked the rest of it's "supply chain" -- perhaps hundreds if not thousands of other inputs necessary to run TSMC and not just doubling down on ASML's cruciality alone.
To be clear, Pat was pretty clear about it from the beginning and the industry experts mostly agreed with the board's push to split up the company.
From stratchery.
> Given this, you can make the case that Gelsinger was never the right person for the job; shortly after he took over I wrote in Intel Problems that the company needed to be split up, but he told me in a 2022 Stratechery Interview that he — and the board — weren’t interested in that:
But where is TSMC going? As far as I can understand it, they would need to rely on geopolitical argument to take customers away from TSMC and I think that's either ultra low volume or a frigid global environment where global trade is winding down.
Yeah only viable way is to be able to compete with TSMC. This requires 10s of billions of dollars of investment (capex amd NRE). This can work if Intel foundry is independent and open to do business with everyone. Apple loves to have a second supplier, nVidia too would love the extra capacity.
The foundry also very much relies on Intel as a customer, and as a semi captive customer who is tolerant of the fab climbing the yield curve using Intel wafers to bring that up. I don't know if the board has thought this through.
They need to get all the talent that left and joined Apple’s silicon team back, IIRC they tried to get Srouji back by offering him the CEO position in 2020 but he stuck with Apple, I have a strong feeling they are probably going to make him the offer again, if he accepts and the Board doesn’t interfere Intel may as well come out of this swinging.
I don't think that they should have TSMC do everything, but being coupled at the hip with the foundry business has not been, and is still not a winning strategy.
If the foundry business could execute (in timelines, in yields, and in sales), then having a vertically integrated company makes a ton of sense. However, that has not been the case for a while and will take years to get them back in a position where they can execute on the foundry side.
Right now the market and the board is dealing with the company equivalent of having your 25 year old kid stay their home rent-free, while spending tons of money, and getting piss drunk, then taking a shit on the hallway rug. At a certain point, you have to kick them out and tell them to get their life together, because the alternative is being homeless.
What they are doing / have done is obviously not working, they aren't being judicious in their spending, they aren't innovating, and everything is being dragged down due to institutional mediocrity.
I don't understand the focus on Intel's board. Intel released a new processor back in October (Arrow Lake) and it was a total flop. That is why Pat Gelsinger was fired.
That's going to be interesting if they intend to sell the fabs. A lot of that money is contingent on Intel not selling them off. Which makes sense: if you take a multi billion dollar subsidy and then just sell the thing that you built with it it's a bit of an obvious money grab.
25 comments
[ 3.0 ms ] story [ 61.0 ms ] threadThere are a lot of questions!
Also, it's very laid back, but Bryan Cantrill's podcast touches on some of these issues. Interesting to see the perspective of an Intel customer [1].
[1] https://oxide.computer/podcasts/oxide-and-friends/2218242
Versus competing with TSMC and potentially ASML?
That's all.
Competing with TSMC using ASML’s kit means beating them on input costs, operations and/or distribution. That’s doable but tougher and more ephemeral than having a technological moat. (It’s Herculean given TSMC’s scale.)
If someone wants to take on TSMC, they should be looking at leapfrogging or sidestepping ASML in a differentiating way. Not out of the gate. But in a PA Semi sort of way.
From stratchery.
> Given this, you can make the case that Gelsinger was never the right person for the job; shortly after he took over I wrote in Intel Problems that the company needed to be split up, but he told me in a 2022 Stratechery Interview that he — and the board — weren’t interested in that:
Where would that funding come from if the product side is a totally separate company?
TSMC took the lead with Apple funding for example.
If the foundry business could execute (in timelines, in yields, and in sales), then having a vertically integrated company makes a ton of sense. However, that has not been the case for a while and will take years to get them back in a position where they can execute on the foundry side.
Right now the market and the board is dealing with the company equivalent of having your 25 year old kid stay their home rent-free, while spending tons of money, and getting piss drunk, then taking a shit on the hallway rug. At a certain point, you have to kick them out and tell them to get their life together, because the alternative is being homeless.
What they are doing / have done is obviously not working, they aren't being judicious in their spending, they aren't innovating, and everything is being dragged down due to institutional mediocrity.