Hacker News new | past | comments | ask | show | jobs | submit login

>The only comparable data point says that this is a terrible idea.

Huh, I would say completely opposite thing. AMD wouldn't have survived if it kept trying to improve their own process instead of going to TSMC.




The problem here is not the success of AMD after splitting, but the complete retreat of Global Foundries from the SOTA process node. If this happens again with an Intel split then we have only TSMC left, off the coast of mainland China in Taiwan, in the middle of a game of thermonuclear tug of war between the West and China.

While Capitalism will likely be part of the solution, through subsidizes for Intel or some other form, it must take a back seat to preventing the scenario described above from becoming reality. We are on the brink of this happening already with so many people suggesting such a split and ignoring what happened to AMD and GF.

The geopolitical ramifications of completely centralizing the only leading process node in such a sensitive area between the world's super powers cannot be understated.

Full disclosure: I'm a shareholder in Intel, TSMC, and AMD.


Let's just hope that if Intel's position is protected because of its strategic importance in the tug of war, it doesn't become another Boeing.


This is a bizarre comparison. Boeing made an entire line of planes that could randomly dive into the ground, and insisted that there be no additional training required for the uptake of those planes. Intel, in contrast, was over-ambitious with 10nm and didn't wait a few more months to incorporate EUV into that process node. The government hasn't banned the use of Intel chips, but the 737 Max 8 was grounded for 20 months. While the pandemic slammed air travel, it has been a major tailwind for the PC and server markets alike.


besides commercial aviation, Boeing (which swallowed Douglas and McDonnell Aircraft, Rocketdyne, etc.) is the second largest defense contractor in the world. It's too critical to be allowed to fail. Intel and Global Foundries have the only nearly state-of-the-art foundries far from China.


I thought Boeing had many issues pre-covid, not just the 737 Max. Starliner immediately springs to mind.


> The geopolitical ramifications of completely centralizing the only leading process node in such a sensitive area between the world's super powers cannot be understated.

This is a fair point but I think it will be less of a concern once TSMC's 5nm fab in Arizona comes online[1]. Samsung is also in the process of building a 5nm fab in Korea[2]. Geopolitically, Korea may still be a minor concern, but much less so than a possible Chinese invasion of Taiwan.

[1]: https://www.anandtech.com/show/15803/tsmc-build-5nm-fab-in-a...

[2]: https://www.bloomberg.com/news/articles/2020-05-21/samsung-t...


I’m pretty sure the Arizona fab is comparatively very very small, although a step in the right direction.


Yes, it says it in the link they cited. The Arizona fab is expected to yield ~20k vs ~100k at one of the Taiwan based fabs.


The price of creating the fabs for a new node increases exponentially with every node. I remember when there were over 20 top node players. Now there are 3 if you aren't counting Intel out. If AMD had remained in the game there's no way they could have won.


I agree with respect to AMD's situation. I think that was the right decision for them then.

I'm saying that there is a difference between the two situations and there are geopolitical factors at play that mean the answer here is not as simple as splitting Intel into a foundry company and a chip design company, due to what we saw happen to AMD's foundry when they split.

I think it's a bit misleading to say that there are 3 top node players right now. Samsung, TSMC, and Intel, while from a business perspective do compete, from a technical perspective TSMC seems to have a fairly significant lead. Like you said, the price increases dramatically every node. If Intel were to split, why would that new foundry company bother investing a huge amount of money in nodes they can't yet produce at volume? Also, Samsung while close to TSMC in competition at this point, still produces an inferior product. There seems to be solid evidence of this in the power consumption comparison of AMD vs NVIDIA top end cards.[1]

My point being, if Intel were to follow the same road as AMD and split up, we could find ourselves in a situation that while better for Intel's business, would arguably leave the world worse off overall by leaving TSMC as the only viable manufacturer for high end chips.

1. https://www.legitreviews.com/amd-radeon-rx-6900-xt-video-car...


I feel like that disclosure isn't warranted since like two of them are in the S&P 500 and everyone here probably has some exposure to that.


Fair enough.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: