This is the type of news that will be relevant in 20 years.
There are only 3 foundries left: Intel, TSMC and Samsung.
If (when) Intel gives up, none of them will have strong roots in the USA. One of the biggest shifts in technological expertise from West to East in history.
Doesn't seem to be obvious that Intel will blink first, before Samsung.
Intel might start making noises to try and get government subsidization though, and I would expect Samsung to do the same to the Korean government.
Having a up-to-date fab is pretty clearly a national security asset, else you are at the mercy of whoever you are buying silicon from. Maybe in the 90s or early 00s when everyone was on the globalization train we would have just let things go, but it's hard to imagine that happening now that hardball realpolitik is back in fashion.
The only chance of this happening realistically is if Intel decide to open their fabs to let others produce their own CPU designs in Intel factories. For example, instead of Apple going to TSMC for their next gen ARM design, Intel decide to do a deal to win that business and produce iPhone CPUs in their fabs. Intel I believe still holds an ARM license from the XScale days, so I suppose an in house design for Android devices could work too.
Historically Intel has been hugely reluctant to do this, as part of the deal when buying Intel chips was that to get access to their advanced fabrication you have to buy their own high margin chip designs. Building other people's chips is traditionally a much smaller margin business, and would be an enormous change to Intel's business model.
I could still see this happening, especially with the current outlook for Intel not looking as rosy as it once was, but it would be a major loss of face for the company that once ruled strong on x86.
The fact that Intel effectively have no market share in the hottest consumer computing market in history (smartphones) is a major failing on Intel's part. Intel selling their ARM business (XScale) to Marvell the year before the first iPhone launched increasingly looks like a pretty terrible decision in hindsight.
> The only chance of this happening realistically is if Intel decide to open their fabs to let others produce their own CPU designs in Intel factories.
Intel Custom Foundry has been a thing since 2010.
That said, it's incredibly low volume compared with the competition.
Intel Custom Foundry historically has not offered the same tech as their internal CPU foundry. Last time we engaged with them their IP offering (memories, serdes etc) was very uncompetitive. It's possible that's now changing but if you are wondering why it's been low volume, that's why.
The more I read about this, the more it seems likely the next big architecture could come from an aliance between Apple and Intel (similar to how PowerPC was owned).
Although Apple are big on ARM, I can’t see them completely abandoning x86 on the desktop. For Apple themselves it’s not that big a deal, but for 3rd party software vendors and users, switching to another architecture barely a decade after the last switch would be painful. On the hardware side Thunderbolt 3 is still Intel only, and clearly Apple are pushing big for that to be the future.
On the other hand mobile has shown the best designs have a mix of processor cores, some optimised for power and others optimised for performance. Intel haven’t shown anything like that for x86 (other than p/c-states), but clearly the same would be beneficial there too.
So why not Intel performance cores, paired with ARM low power cores? Apple have the low power design know how, and Intel have the fab know how. Both companies have a lot they could benefit from this partnership. Maybe more importantly, from a political side it would be an “all American” chip.
There are plenty of fabs left. The problem is that each iteration comes with a staggering cost that takes time to make back. A lot of plants quit chasing the latest and went to niche markets where the money is better. That said, if some of the remaining competitors drop out, and prices rise enough, you'd see a couple reemerge.
There are only 3 foundries left: Intel, TSMC and Samsung.
If (when) Intel gives up, none of them will have strong roots in the USA. One of the biggest shifts in technological expertise from West to East in history.