If you don't build factories unless you are sure of demand, you are already out of the game.
You need to lead and technological leadership creates demand. If you wait for the orders before building factories, you will get no orders, because the competition already took those orders.
RTXMonster
Reputable
- Jan 23, 2021
- 3
- 0
- 4,510
Man I feel Lip Poo might work for someone else on the sly. The moment he has stepped in he has done everything in his power and decision making to make sure the foundry side never even gets started. Firing people they need to get it going is just genius 👏
I wonder what these expectations are of his? Unreasonable? Unobtainable? I mean I don't expect them to just jump out the gate taking market share from TSMC. That is a very stupid way of thinking.
Everything this tart does is good for nearsighted investors but in the long run I don't see how all these things don't destroy Intel. Gelsinger at least had a better path for them to be relevant again.
The government better get their Chip Act money back at this point. This whole thing stinks to no end.
I wonder what these expectations are of his? Unreasonable? Unobtainable? I mean I don't expect them to just jump out the gate taking market share from TSMC. That is a very stupid way of thinking.
Everything this tart does is good for nearsighted investors but in the long run I don't see how all these things don't destroy Intel. Gelsinger at least had a better path for them to be relevant again.
The government better get their Chip Act money back at this point. This whole thing stinks to no end.
thestryker
Judicious
- Apr 19, 2016
- 5,832
- 4,019
- 36,690
This is an uninformed take as to the state of the industry and packaging. CoWoS isn't some super advanced packaging technology that nobody else can match, but rather a good high volume packaging technology. Intel has advanced packaging technologies which match the capability of TSMC's some of which are better and some that are worse. What Intel hasn't had, until last year, is enough volume to seriously offer them to external customers. Samsung also has their own packaging technologies, but from my understanding doesn't match all of the capabilities of the other two.
The US governments only chance is nationalizing Intel, putting 100 billion in investment and changing the board. Not going to happen. Nobody else is going to invest in building leading edge Fabs in the USA.
thestryker
Judicious
- Apr 19, 2016
- 5,832
- 4,019
- 36,690
This... doesn't really mean a whole lot. It's a misrepresentation of reality (which is something CEOs do all the time) to push a specific narrative. While 18A was the first advanced node being pushed for external customers it's also introducing GAAFET and BSPDN the latter being the first industry implementation. These things limit how much external input is plausible. That's one of the benefits of TSMC taking things in a more step by step fashion rather than swinging for the fences.
Intel needs 18A to be good and draw in customers, but they didn't need it to be huge out of the gate. They also need capacity so they can produce their TSV nodes both of which are things they've been missing.
In business terms Intel is "shrinking into oblivion".
I saw this with my old company, Xerox, who is now a shadow of what it once was.
TerryLaze
Titan
- May 9, 2016
- 10,906
- 3,102
- 76,440
Huh?! Am I not understanding you here?! Even if these things make 18A less suitable for external customers, the quotes are about 14A and how they will work with external customers from the get go to avoid these things.
It might need a radical redesign if these things in deed need to go but that's probably why he's willing to drop it if there is not enough guaranteed demand.
Also are you just assuming that these are the things that make 18A less suitable for external use or do you know anything?
TerryLaze
Titan
- May 9, 2016
- 10,906
- 3,102
- 76,440
If they were after a bail out they wouldn't be trying to lessen the financial blow by selling un-needed things and firing people, they would just pile it all on to make the losses as big as possible.
Cores cost money per core. If anyone thinks these next gen cpus are going to be same price are cheaper, please pass that pipe i need some dreams fulfilled too.
18a supposedly cost 2.5 times more to produce the same wafer. 52 cores to beat 24 man this is some bad logic.
18a supposedly cost 2.5 times more to produce the same wafer. 52 cores to beat 24 man this is some bad logic.
>The only way I can see this turning around for Intel (as a manufacturing company) is if this is a bid to get the US government to bail them out in a big way.
I'm reminded of the advice given to me years ago, that there aren't 4Ps in marketing, but 5, of which the last is public policy.
Public policy does loom large here, not just for Intel, but for practically every US and international company doing business in the US. Trump's tariffs are impacting every industry, along with the huge uncertainty they entail. Every company is holding off on big investments until the smoke clears. This factor alone is enough for Intel to be conservative on its fab investments.
...not least of which is the upcoming semiconductor sectoral tariff that was promised a few weeks ago. Maybe it will be announced on Aug.1.
[BTW, per info from the few trade agreements thus far, we're looking at a baseline tariff of 15% across the board, and higher for countries that won't bend the knee.]
As far as bailout for Intel, it does have some fairly solid leverage to ply, since its fabs would be a lynchpin in Trump's "AI supremacy" plans. The flip side of that is that Lip Bu Tan has not hobnobbed with Trump--has not "contributed" money--unlike many other tech CEOs. He didn't kiss the ring. When all policies revolve around the whims of one man, personal relationships matter.
Intel received $2.2B of the $7.86B awarded under the CHIPS Act. Its pullback in fab construction will likely have impact on the rest of the money. That, and again it depends on Trump's whims as to whether he will honor agreements from the previous US admin.
Summing all these up, and the answer is a big "who knows." It's the same answer every business has faced since April 2.
>But I don't think even that would work, as the problems appear to be systemic with Intel's foundry side.
That's what Tan is setting about changing, the "business as usual" attiude that permeated Intel's MO thus far. Fab construction is a biz with long lead times, and many aspects of it are inherited. But 14A is sufficient far enough out for Tan to set his own imprint. We'll see if he succeeds. I'm rooting for the guy (with actual money.)
I'm reminded of the advice given to me years ago, that there aren't 4Ps in marketing, but 5, of which the last is public policy.
Public policy does loom large here, not just for Intel, but for practically every US and international company doing business in the US. Trump's tariffs are impacting every industry, along with the huge uncertainty they entail. Every company is holding off on big investments until the smoke clears. This factor alone is enough for Intel to be conservative on its fab investments.
...not least of which is the upcoming semiconductor sectoral tariff that was promised a few weeks ago. Maybe it will be announced on Aug.1.
[BTW, per info from the few trade agreements thus far, we're looking at a baseline tariff of 15% across the board, and higher for countries that won't bend the knee.]
As far as bailout for Intel, it does have some fairly solid leverage to ply, since its fabs would be a lynchpin in Trump's "AI supremacy" plans. The flip side of that is that Lip Bu Tan has not hobnobbed with Trump--has not "contributed" money--unlike many other tech CEOs. He didn't kiss the ring. When all policies revolve around the whims of one man, personal relationships matter.
Intel received $2.2B of the $7.86B awarded under the CHIPS Act. Its pullback in fab construction will likely have impact on the rest of the money. That, and again it depends on Trump's whims as to whether he will honor agreements from the previous US admin.
Summing all these up, and the answer is a big "who knows." It's the same answer every business has faced since April 2.
>But I don't think even that would work, as the problems appear to be systemic with Intel's foundry side.
That's what Tan is setting about changing, the "business as usual" attiude that permeated Intel's MO thus far. Fab construction is a biz with long lead times, and many aspects of it are inherited. But 14A is sufficient far enough out for Tan to set his own imprint. We'll see if he succeeds. I'm rooting for the guy (with actual money.)
thestryker
Judicious
- Apr 19, 2016
- 5,832
- 4,019
- 36,690
You obviously didn't understand me at all. It's a too many cooks in the kitchen type of situation where the more Intel tried to do with 18A the more difficult making a successful node would be. This means that the opportunity for external input would inherently need to be limited. 14A shouldn't have this problem at all because all of their potential customers already have the 18A PDKs and can provide useful input as to what can be changed to better address their use cases. I'd imagine to some extent this feedback will have shaped 18A-P as well.
TerryLaze
Titan
- May 9, 2016
- 10,906
- 3,102
- 76,440
Then what was this part about?!
"This... doesn't really mean a whole lot. It's a misrepresentation of reality (which is something CEOs do all the time) to push a specific narrative. "
If 14A can possibly be much more relevant for external customers then the quotes he did are good.
- Jul 26, 2008
- 4,556
- 1,172
- 24,840
Not remotely and nothing indicates as much.
Intel is in a similar situation that AMD was in, only Intel is way more healthy due it's Enterprise products. This is more a question of future strategy than eminent survival.
They could dump their entire foundry side and still continue to exist, they would be smaller but not dead.
thestryker
Judicious
- Apr 19, 2016
- 5,832
- 4,019
- 36,690
His implication that Intel simply chose not to work with external partners on 18A rather than having actual reasons to not. Only putting forth part of the story to make the current plans look like a major departure from the path the company was on.
It's kind of like how Gelsinger loved to go on about Larrabee and what a missed opportunity it was. He's right Intel should have gone all in on compute, but Larrabee was very unlikely to be the right way forward.
Last edited:
I read this message as either
1) Management knows that they will fail and hinting government to continue handing money on it for years to come. Then if they still fail they hope that either Emir or donkey will be in a different reality and the management still get their golden parachute
2) This is a hint that some largest investors know that the big problems with Intel are inevitable and have a short order on it
1) Management knows that they will fail and hinting government to continue handing money on it for years to come. Then if they still fail they hope that either Emir or donkey will be in a different reality and the management still get their golden parachute
2) This is a hint that some largest investors know that the big problems with Intel are inevitable and have a short order on it
ikjadoon
Distinguished
- Feb 25, 2006
- 1,998
- 62
- 19,860
This claim simply does not compute.
Are Samsung & TSMC building brand-new fabs purely on the hope customers will come? No. They waited until current fabs were near capacity, confirmed demand with signed contracts, and then decided to expand.
Intel's fabs are empty of external customers. Intel's own inventory has a major glut.
//
To make this plain and obvious: we should expect zero external customers means maximum one (1) external fab, ready to be used. Not more than one: just one. Why build 2, 3, 4, 5, 6 fabs when you can't even fill one?
Once that first external fab reaches peak capacity and Intel gets confirmed more demand, then Intel expands into a 2nd external fab.
See, no company will send Intel a "huge" order on day 1. It doesn't make sense: external customers want to de-risk, too. Likely any CEO that would send the company's flagship GPU / AI accelerator / CPU to Intel Foundry exclusively would be lambasted and humiliated; it's just too risky when there is no proof Intel can handle it.
//
For gamers here, it's like taking a loan to buy an RTX 5090 with the hope you might become a professional CS:GO player one day and win enough money back in tournaments to pay for that GPU.
Doesn't make sense. Buy within your means, get good, and then expand when you have the necessity.
ikjadoon
Distinguished
- Feb 25, 2006
- 1,998
- 62
- 19,860
This is a 10-Q filing. Every company writes down every possible risk to avoid getting sued later on that they didn't disclose a risk they were aware of.
This section is not written by management nor bean-counters nor even the CEO. This "Risk Factors" is written by lawyers. It's likely someone, somewhere at Intel admitted to a Intel's in-house counsel, "Financially, without external customers, Intel Foundry can't invest after 14A because Intel will not have enough money for 10A or 7A after 14A's R&D costs."
And Intel's lawyer likely replied, "Well, if that is so, I am required by law to admit that as a risk factor in the 10-Q."
External foundries must generate profits to build future fabs.
How else can Intel run the Foundry? Who is going to keep throwing money at Intel? Just look at its stock: the market has not believed in Intel's profit-making abilities for two years already.
This isn't like Intel 22nm or 14nm, where Intel's own products are so popular & so profitable, the entire Foundry is sustained off CPU sales. That is completely impossible at 18A / 14A because of how damn expensive the R&D and equipment and labor and fabs are.
^^ This is what many are missing. Intel needed external customers to continue its Foundry.
No foundry on Earth can ever ship a ~1nm node without massive customer contracts. Nobody has that much money "lying around".
//
That this risk factor is new this quarter implies Intel is admitting it vastly overspent on fabs that still have no customers. Do we get that? Imagine TSMC building fabs that nobody was using. People would think TSMC is putting themselves at unbelievbable risk.
It's already too late for 18A to recoup the costs because external customers never showed up.
To think about it with random numbers:.
18A internal: 1 fab = cost $10 billion, revenue of $12 billion
18A external: 3 fabs = cost $30 billion, revenue of $0
$2 billion in profit is not enough to build another fab.
$2 billion in profit is not enough to develop 14A.
$2 billion in profit is not enough to sustain the Foundry.
These are made-up numbers, but this is the calculus (see below for the real numbers).
How can Intel keep going like this? Fabs aren't a one-time cost: think labor, maintenance, raw materials, utilities (power, water), and God forbid, retooling for the next node.
//
Intel must find external customers and quickly. It just hasn't happened yet and Intel did not save enough to go on this fab spending spree without ever even filling up one fab with external customers.
I'm not sure if people truly get how precarious Intel's financial position is, per the 10-Q:
- Intel has $51 billion in debt.
- Intel has $9 billion in cash on hand.
- Intel is losing $2 to 3 billion per quarter.
- Intel has $109 billion worth of plants & property—with zero external customers.
Intel needed to fill 18A fabs to the brim to put that money towards 14A, 10A, and 7A. Intel will need to use 18A for the next 5-10 years, just like TSMC, just to recoup the cost of 18A, much less fund future nodes.
^^ this is why Gelsinger said "We bet the company on 18A."
abufrejoval
Honorable
- Jun 19, 2020
- 1,011
- 643
- 12,060
Well that's pretty much what China has been doing from a much less advantageouos position in terms of technology.
But they have a huge and still growing domestic market as well as plenty of other potentials, who'd prefer China's wares over what the US has to offer.
But Intel, Microsoft and the MAGAlaniancs all are doing their best to piss off not just their domestic customer base, but pretty near everyone else on the planet.
And that means that the better technological starting position isn't worth a dime in bailouts.
abufrejoval
Honorable
- Jun 19, 2020
- 1,011
- 643
- 12,060
There is a slight difference in size. And AMD found a buyer who was ready to invest.
And that buyer didn't just let them dump it either, AMD was tied to Global Foundries almost to the point of choking
Even in the US, where dumping corporate liabilities on whoever happens to live close by has been the preferred choice since the Wild West was grabbed, disbanding the fabs and the people involved may not be that easy or cheap, while I'd rather not think about the piles of toxic materials left behind.
From a belly button perspective all that may look simple, but at sufficient size or magnitude, even light gets bended by gravity.
CHIPS act funds are not disbursed without hitting the milestones: https://www.oig.doc.gov/wp-content/OIGPublications/OIG-25-021-I.pdf
TRENDING THREADS
-
Question Can someone label this photo showing where all the components are?
- Started by nelska
- Replies: 5
-
-
-
-
Question Why £1000 for a motherboard and 128GB of ram
- Started by Hasselblad
- Replies: 10
-
Discussion What's your favourite video game you've been playing?- Started by amdfangirl
- Replies: 4K
Facebook
Twitter
Instagram