87 comments

[ 3.1 ms ] story [ 161 ms ] thread
Intel has a clear strategy of becoming a big player in the semiconductor industry in the next decade. I honestly think it will work. Many people have observed that it is a cash-intensive industry with tiny margins, but once these huge investments have been made and there are only two or three companies that can dominate the market for five to ten years, you can extract huge value before any competitor catches up.
This comment could have been written 50 years ago.
Compare the number of players with leading edge fabs then and now though.
There were no players without fabs back then.
Yup, the major change has been the capital to get new fabs off the ground has become unimaginable. TSMC is practically the only game in town for chip manufactures ATM, everyone else that will fab 3rd party designs is way behind.

GlobalFoundries state of the art, for example, is a 12nm process. TSMC sits at 5nm, about 4 generations ahead.

There are trillions of dollars in capital floating around looking for a better place to park it than treasuries. Major infrastructural projects like semi fabs should be a no brainer - the risk is not that high due to evergrowing demand and the trust that governments won't let an Intel or TSMC fail.

Even just look at the VC landscape - there was $621 billion spent in venture capital globally last year, which is one of the worst ways to invest your cash.

Point being there's plenty of capital out there for giant projects like building fabs, I just don't think it's sexy enough yet or the financiers haven't taken a deep enough look while governments are expected to make the first moves.

Samsung is not that much behind, recent nVidia GPUs and Qualcomm CPUs are manufactured by Samsung (8N and 4nm, whatever those mean anyway). While density and perf/watt is worse than TSMC, it's THE competitor for TSMC.
And they still would have been right?
> becoming a big player

Are you saying they aren't a big player now?

Gartner puts them at Nr. 2 in 2021 after Samsung Electronics. However it should be noted that for Intel this seems to include total Revenue while for others it seems to be just their pure semiconductor revenue. If Intel were to sell their chip manufacturing as a subsidiary to themselves at market price, that subsidiary probably would be way further down the list.
Wouldn't this require them to start making SOCs for mobile phones?
In other words producing for chip designers such as Apple, Qualcomm, or Samsung.
US-based TSMC.

They have a good chance of succeeding. They are likely to get a lot of government help.

They’ll probably start with older systems, as there’s already a huge need for these, and they have a gazillion existing fabs. It won’t compete with their current lines, and means they don’t need to mothball their legacy fabs.

> They are likely to get a lot of government help.

Indeed -- the not so subtle subtext of their pitch is "China, eh? Have fun with that.". So a pitch based on the idea that it's strategically important to bring that chip manufacturing "home".

Which is why they picked Israel. /s

> Which is why they picked Israel.

Well, there's a lot of history in that relationship.

Israel really does have some of the finest scientists and engineers in the world -partly thanks to the antisemitism of a lot of other nations.

I have a feeling that Israel (and Iran) have the potential to completely transform the Mideast, but all that fighting is keeping that from happening. If they ever started to work in cooperation with some of their neighbors, we'd see a new power bloc emerge.

If they ever actually have a real, lasting peace, out there, the rest of the world had better hang on, as every single one of those nations would love to not have to be dependent on some superpower.

Yes. I guess what I meant is that there could be an element of the strategic in the choice, particularly for a business that seeks government money in the near future if the US House of Representatives does swing back to GOP control.

Though Tower Semi are clearly enormously capable, so it's hardly a bad acquisition, even absent seeking access to a pot of money with a strategic aspect to it.

> I have a feeling that Israel (and Iran) have the potential to completely transform the Mideast, but all that fighting is keeping that from happening. If they ever started to work in cooperation with some of their neighbors, we'd see a new power bloc emerge.

That's exactly what is happening. The emerging power block in the Middle East is Israel+Arab States vs Iran with Turkey playing its own game.

https://en.wikipedia.org/wiki/Abraham_Accords

An Ayatollah free Iran has huge potential, very true. There are reasons for cautious optimism. None of this has anything to do with Intel/Tower however.

The Intel presence in Israel goes back generations and Tower is a cheap and obvious pickup for them.

>Well, there's a lot of history in that relationship.

The Israel branch basically saved Intel when NetBurst failed completely - at the time Intel was behind as much as today.

Intel Foundry Services (IFS) is marketing themselves as the only fab capable of producing all three of x86, Arm, and RISC-V.

Back when Intel had a process advantage over everybody else I guess it made strategic sense for them to not let others use their process. Now when that process lead is gone, refusing to fab chips for non-x86 ISA's just means those customers will go elsewhere.

> Intel Foundry Services (IFS) is marketing themselves as the only fab capable of producing all three of x86, Arm, and RISC-V.

Capable? From the fab point of view, all three of those are just rectangles.

I mean, same rectangles won't work for different foundries. It is saying we have these IP cores tested and working with the process.
This, and also the licensing question. Intel has said they will license x86 cores for use in 3rd party SoC's/chiplets, but maybe only if you use IFS.
That sounds a lot like a tying arrangement, which could be unlawful under US anti-trust laws.
> Intel Foundry Services (IFS) is marketing themselves as the only fab capable of producing all three of x86, Arm, and RISC-V.

TSMC is producing all of these. Not like the architecture matters from a foundry POV anyways.

TSMC is producing some x86 cpu's for Intel. I think what they're saying here is that they will license x86 cores for integrating into 3rd party SoC's/chiplets, but only if you use IFS.
They're also producing all x86 CPUs for AMD.
(comment deleted)
Yes [1]? The argument is about whether Intel refuses to license their processor designs to 3rd parties to be fabbed in any random fab.

[1] Evidently the x86 cross licensing deals between AMD and Intel doesn't preclude AMD from using 3rd party fabs.

(comment deleted)
>f becoming a big player in the semiconductor industry in the next decade

Did you mean

"a big player in the semiconductor _foundry_ industry in the next decade" ?

Intel has been the biggest semi player for most years during the last three decades.

Where do you get that the profit margins are tiny? TSMC has a net profit margin of 38%; that’s higher than Google (27%).
They don't have a strategy, they react to their downfall and are rushing to find a solution

They are lucky the pandemic happened, it gave them the opportunity to take a step back and realize how screwed they are

(comment deleted)
When reading headline I was thinking about Intel removing its competitor. Then I searched a bit and found this: “ Tower Semiconductor Ltd. is an Israeli company that manufactures integrated circuits using specialty process technologies, including SiGe, BiCMOS, SOI, mixed-signal and RFCMOS, CMOS image sensors, non-imaging sensors, power management, and non-volatile memory as well as MEMS capabilities.”

Looks like that Intel wants to go deeper in foundry business offering more services. I can speculate, that Intel has very good silicon process for logic circuits and will improve their offerings with technology from this acquisition. Sounds like a reasonable strategy to me.

So it is some kind of complement to Intel's business? By the price Intel is paying compared to Tower's revenue, it really sounds like Intel is paying a lot without being able to add anything?!
It's somewhat complementary to their Israeli fab business.

They may also think having a multibillion dollar fab investment next to the Golan Heights is safer than having one next to the Gaza strip. ¯\_(ツ)_/¯

Yeah, yeah. A bit to chat about when Pat Gelsinger roams about thrashing TSMC for having their fabs at an insecure location.
Kinda spooky how much of the cutting edge semiconductor fabs are in disputed territories...
they are purposely built their for Geo-political reasons, to ensure those area's stay "disputed"
Or the disputed places make unusual investments due to their status?
I disagree. Israel is unique in terms of it's human capital. Tech companies have locations there for the same reasons they have locations in Oxford and Cambridge in the UK. There are 135 per 10,000 people employed as engineers there, compared to the next closest which is the US at 70 per 10k. That would already be a unique draw for Intel, if it weren't for the original reason they set up shop there; Dov Frohman. He was the first GM at Intel (also the inventor of EPROM), and wanted to relocate there to accept a Professorship. They didn't want to lose him, so they agreed to let him open an R&D lab there. This ended up being a huge success (they designed the 8087 there) and they continued to grow on that success, to the point where Intel is now the largest private employer in Israel.

I recommend Asianometry's video essay on Intel's success in Israel: https://www.youtube.com/watch?v=3lEgJQgAw3g&t=894s

He also covers the rise of such silicon manufacturing in Taiwan - which I also disagree on it being geo-politically motivated, with it being successful long before it's national identity came under threat from mainland China.

I think the issue of "how to reunite the rogue republic of Taiwan with the motherland" predates silicon manufactoring in Taiwan by a few years :-D
The "geopolitics" is played up as the US now needs an excuse for aggressive industrial policies aiming to move jobs from "allieds" such as South Korea and Taiwan and back into the US.

Pat Gelsinger panders to this as he eyes a windfall for his business, so much that it is getting harsh reactions from Taiwan:

Intel CEO: Taiwan Isn’t a Stable Place. TSMC Founder Has Other Concerns

https://techtaiwan.com/20211207/gelsinger-intel-taiwan/

Completely absurd comment. Israel is a minnow in the semiconductor manufacturing industry and many fabs are located in Taiwan because Taiwanese people invented the foundry business.
that area is not considered disputed at all. there is a difference between insecure and disputed or maybe I'm getting it wrong?
(comment deleted)
There's a lot more water in the Golan Heights, which is really why Israel decided to take it over in the first place.
Water is crucially important to fabs, true. But get real, water is important for everything in Israel, and why it had strategic value to warrant annexing it after the whole--the whole war that happened, the casus bellum of which cannot be done justice in short comments like this.
The parent said Israel took over Golan Heights for its water, not "...for its water for fabs".
Fair. But my argument is with it constituting encroachment, like Syria in no way brought that territorial loss upon itself in any capacity.

My objection was one-sided denigration of Israel when it appears--in the comment I replied to, it was mild and implied--which mitigates things. Can't help notice Anti-Israel speech as gotten to be very common on HN in the last two years. If you want to criticize or even denounce Israel, this forum is for sure an appropriate time and place for that, and it appears nobody will claim anti-Semitism in the 2020's like they would in the past (as far as I can tell, this decade only American Jews are off limits, Israel is fair game).

But historical falsifications are where I draw the line. I claim that is incredibly dangerous, leads people to believe that a group--here, it's Israeli Jews--is morally inferior. Do you know what happened next in the Middle Ages, or the Renaissance, or the Enlightenment, or even in the 19th century, any time the stereotype of moral inferiority took hold?

That is precisely the moment to "speak for the Jews," as the poem put it.

Israel is so small that according to your definition of "next" everything has to be "next" to the Gaza Strip, the Golan Heights and the West Bank, and also to Lebanon, Syria Egypt and Jordan for good measure.

Seriously, Tower is far away from the Golan Heights in the Israeli sense. It is however in the Galilee (big Arab population) and "near" the West Bank.

I think the closeness to the Technion university and to Intel's partners factories (KLA Temcor...) is a much more significant factoid

> By the price Intel is paying compared to Tower's revenue

~3.5x sales is not a lot, especially not in this market (either broadly or semis specifically).

AMD is at 9x sales; Nvidia is 27x; Analog Devices is 11x; Texas Instruments is 8x; Applied Materials is 5x; Micron is 3.6x; Taiwan Semi is ~11x; ASML is ~13x; KLA is 7x; LAM is 5x - and so on.

These companies also don’t all have the same margins, so benchmarking on a sales multiple doesn’t give a good feel for valuation. Everyone does it anyway though, so in that sense it works.
AMD isn't a comparable business for this list, because it no longer has fab facilities. That was spun off to GlobalFoundries.
These technologies are key to certain high speed applications. As we move from monolithic devices to MCMs with lots of high-speed IO these technologies will be important.
Tower is one of the most capable CMOS image sensor fabs. They're generally used by MFT manufacturers for their image sensors.

The other, and the biggest player in this image sensor business is Sony.

This is a completely different lithography and technology when compared to CPUs and GPUs (larger feature sizes, different requirements, etc.).

Intel's transforming itself to be a fab operator it seems.

Scooped you by 21 hours: https://news.ycombinator.com/item?id=30344944

I think this comment will be downvoted, but just want to run the experiment to be sure. I'll delete if so .. No worries.

You'll be downvoted for talking about downvoting (just as I will be). Just avoid.
No I think it's helpful in these cases where I'm just trying to learn about the community.

Heh, we probably ought to avoid continuing to talk about and make a thread about it here :)

But maybe there's some useful distinctions to be made. I think the times where's it's bad is if I go, "Don't downvote..." or "Downvotes are idiots" or whatever...Here I'm just honestly being like, I think this will be a "low value" comment and I'm happy to remove it, but I'm not entirely sure so I want to check. I think that's useful to make it clear and transparent.

Actually I'm surprised it got upvoted. I thought people would take issue with me saying "I was there first"...but anyway, I learned something. Yay :)

I downvoted you because I do think your comment does add little to the discussion other than "I posted it first!", which isn't all that interesting all things considered. It just seems bitter.
That's what I thought people would feel, and why I said I thought it would be downvoted. But thanks for letting me that's what you think anyway!

I don't like to admit it because of how silly and weak I think it looks, but, yep I did feel it was unfair and I was showing a little bitterness there--correctly identified, ginko!--I felt sad about it, like "why no upvotes for me?"--but also posted it with a goodnatured thing too, like "I scooped you! But you got the votes!" which can always be a dynamic with news scoops. It was both, you know. And I don't really think it's unfair, because it's basically random how a post gets up. Plus, Haaretz has a paywall, maybe that contributes too...But you always wonder, you know...heh :) Thanks for adding to my shame there, ginko :<...

edit: Oh, I see you got downvoted. Poor you. That's very meta.

edit 2: Ooh, I see the original post has been flagged. I definitely think we should stop talking about it now, heh! OMG let's stop please! :)

You're not exactly Babe Ruth calling the shot on this one. If you post something that isn't intellectually gratifying and purely petty, then yes, you're going to be downvoted on Hacker News.

If you posted something along the lines of "Here's the version that isn't paywalled" or "this article has a more interesting analysis", people wouldn't have dogpiled you. Instead, you buried the lede and mentioned some pseudo-experiment, which isn't going to get a lot of traction around here. If you want to endlessly fight with people over attribution credits and updoot points, you'd probably get a real kick out of Reddit.

(comment deleted)
(comment deleted)
Interesting...

"Babe Ruth's called shot."

Heh, no not in this case :). I thought it was going to get downvoted quickly, but it didn't. edit: I see that finally it has hit 0, after many hours. So, I guess I was, similar to, Babe Ruth calling the shot, here. Heh :)

It was a real experiment, not pseudo.

Agree about the intellectually gratifying, and your alternative suggestions are good. I was considering something like that too (thought I posted in a reply, but I edited it out).

It wasn't purely petty, as I said, sportsmanlike. Is feeling something unfair, or feeling sad about something, petty? I don't think so, I think it's just normal...but like I said, I feel it's embarrassing to admit in this case, because it seems silly to me. But I guess...well I do care about it, so I guess it can't be that silly.

It wasn't fighting...not at all.

And I don't think people 'dogpiled' -- just sharing their perspective and/or trying to teach me something. And I learned something. So that's good. 'Dogpiled' is interesting, haven't heard that term before...

Wouldn't say I buried the lede, because there were only 3 parts to the post: link, scoop and experiment, and enough people saw it to get it many more upvotes...

Heh, you do seem to find a lot to disagree about, misrepresent, or take issue and criticize there (not sure exactly how to describe what you do above)...could it maybe be you who was looking to fight? I honestly don't feel like that, so please let's not continue. Peace, please? I would love that :)

Actually I made that mistake I sometimes make.. Because of high empathy, I will take on someone else's perspective and agree with them because I can feel how they see it like they do... Even when it's not actually like that for me.

Yesterday I wasn't bitter at all...I think I just 'admitted to it' because I didn't want to disagree. I wanted to try to find like a common ground and because of my empathy with you, I could really feel your perspective. But actually I wasn't bitter, I was just curious that this one got up and mine didn't and I was also competitive like, "you know, I win anyway cuz I scooped you. Okay you got the votes but I scooped you, so I win" That's honestly how I felt about it. So for me, it wasn't bitter at all, just curious and competitive. I do get if you feel that way about it, and I get if you felt there was bitterness there or you just thought that's how you would feel in that situation, but I think you were just finding there what you bring to it. So thank you for sharing but that was just you projecting your feeling onto mine, and then me taking that on, as a mistake. :)

In future I hope I can say these type of things more concisely. I feel it's annoying when people misrepresent me, and then I take longer to represent myself clearly, than they took to misrepresent me. I would love to make a shorter reply than the original misrepresentation! That's a goal I have, it's a definite opportunity for improvement for me.

Anyway I'm sorry to me for just agreeing yesterday without really thinking through and being authentic to me...even though I kind of I felt what I was saying was wrong for me at the time, but I just decided to find a way to agree, that was a mistake. It's a mistake I've made sometimes, because of high empathy and not always happy to disagree. I'm sorry to everyone else here is sort of you know reading all this stuff but it's important enough both to say so people know me authentically here but also is a lesson for me that I want to post this. Thanks!

Since you’re conducting an experiment, I’ll provide my feedback. Usually I’d downvote a simple comment talking about being first to post but you also provided a link to the previous discussion which I thought was worth upvoting.

Regarding the talk of downvotes, I’d usually downvote for all the classic reasons but you turned it into an experiment which will spark discussion, so that’s an extra reason to upvote.

I take my upvote/downvote powers very seriously. Or maybe a bit too seriously?

Edit: after making this comment I clicked and saw your post got no comments, so I must now consider it to be a “simple” comment with little value.

Hmmmm. I guess I’ll just take back my upvote but refrain from downvoting.

Hehe, thank you for the comprehensive thinking portrayal. I really like that. I also take the down/upvotes thing seriously. And when I first got the power really didn't want to use it to down people even if they were really mean or I really disagreed. I try to upvote people who add information or a good feeling, but I still rarely downvote. At first I rarely upvoted as well...but now a lot more.
Is this Intel’s PASemi moment or something entirely different?

Remember PASemi were the folks Apple bought that gave us the A series chips and made Apple the chip powerhouse it is today.

I think it's entirely different. Tower is a specialty foundry.
> manufactures integrated circuits using specialty process technologies, including SiGe, BiCMOS, SOI, mixed-signal and RFCMOS, CMOS image sensors, non-imaging sensors, power management, and non-volatile memory

Aren't these the kinds of technologies used to produce the kinds of chips that are in great shortage atm?

Or perhaps a better way of stating what I'm trying to say is... aren't these the kinds of technologies that could be used to produce chips that can replace those that are in great shortage atm?

Not as dependent on highly advanced lithography machinery from the likes of ASML, but not so ancient as those that have recently stopped churning out the same basic chips after 20-30 years.

Yes. On the investor call one of the points was that Intel's cash would allow tower to expand and modernize its old 150mm and 200mm fabs, and to expand PMIC production enough to be competitive in that market.
Intel will succeed as long as it isn’t grossly mismanaged simply because the US government needs them to succeed. I’m also surprised they don’t pressure Apple to start building fabs in the US as well considering they own the entire stack now.
Apple doesn't fab chips itself; those come from TSMC… who are building fabs in the US now.
I kind of expect Apple to start fabbing chips eventually.
I'd foresee Apple ordering chips from Intel Foundry Services' new US fabs before Apple actually building their own fabs. Although AFAIK, IFS fabs do not yet compete in the same nodes as TSMC.
(comment deleted)