What I don't get is that there are at least a couple of very large, very valuable US companies that need to have their CPUs/GPUs fabbed (Apple and NVidia) and are currently dependent on fabs in Taiwan, a geopolitically risky place to be that dependent on. Both are sitting on huge reserves of ca$h. Why not either outright buy or buy a large stake in Intel to recapitalize it and allow it to finish the new SOTA fabs it was building? The CHIPS act was intended to help the likes of Intel and Micron, but the current admin has apparently blocked any further funding. If the current admin was serious about US semiconductor manufacturing it would try to arrange some kind of shotgun wedding where Apple & Nvidia (and others) take a stake in Intel to keep it afloat. Perhaps some kind of a consortium where the investing companies get priority in getting their parts fabbed? There's really no other US alternative for advanced semiconductor fabrication unless you're going to start from scratch and that doesn't seem like a viable idea.
Yes, I understand the argument that Intel management screwed up for too long and this is the market at work, but that ignores the geopolitical risks of what we're going to end up with. Forming some kind of consortium to keep Intel fabs running (and new ones built) could also include completely changing the management of the company.
To me, Apple and NVidia have already voted "no" to major Intel investment. This is evident from how large is their order book with TSMC and Samsung. I think the newest TSMC fab in Arizona is one step back from the best in Taiwan, and I think they will build another fab in Arizona soon (if not already started).
Is it because the employees themselves are so incompetent that no one wants to take this burden on. Besides, TSMC is expanding in Arizona and Samsung is expanding in Texas.
> Is it because the employees themselves are so incompetent that no one wants to take this burden on.
I don't buy this. I think the primary problem was mismanagement especially in the 2008 to 2020 timeframe. Too many bean counter CEOs during that period who did not understand the need to constantly invest in SOTA fabs.
> Too many bean counter CEOs during that period who did not understand the need to constantly invest in SOTA fabs.
I am not here to defend Intel, but I don't think this is the correct interpretation of events. Basically, Intel failed in their fab process R&D to keep up with TSMC and Samsung, and that is not lack of effort or money. Since their fab process R&D was going so poorly, Intel slowed down their fab construction rate. This makes good business sense to me. The truth appears to be that Intel fab process got beat fair and square by TSMC and Samsung.
Intel absolutely flubbed some nodes and bad employee execution was a part of it.
But management has consistently tried to tell customers what they want/need. Intel has a history of developing products with no customer base or pulling out of markets too early. Neither of those are the responsibility of low level employees. That's higher management.
One of the big concerns about the GPU division is "Will Intel keep going long enough for this to matter or will they pull out the second there's an issue?"
Because it was incentive based. Intel has to hit milestones to get funding. Everything Intel does takes years.
Intel is also likely going to lose hundreds of millions in incentives from Oregon for failure to meet hiring objectives, but they have a while to do that.
As I understand, the best fab tech is TSMC (Taiwan) and Samsung (Korea). Do you really expect China can surpass both in only two years? It seems unlikely, as they don't have access to high-end fab equipment from ASML.
IMO you’re misreading: as I understand it China is poised to take Taiwan in 2027. It’s my understanding they aren’t being shy about announcing it. It ceases to be about tech and becomes more about who controls the state.
Edited to add: this was not the point being made, I am aware. Just my thoughts on the matter.
