Re:Intel let the chips fall where they might
by Panzer04 on Monday August 01, 2022 @09:09AM
(#32305420)
I’m always slightly confused at how easily intel is written off. They still generate more revenue than AMD and Nvidia combined, they have a huge foundry business attached, their processors are competitive and they are making the necessary investments to catch up.
AMD and Nvidia will always be beholden to TSMC - yes, they avoid the capital outlay and risk of foundries, but they also give up the possible upside of owning the foundry.
Perhaps intel will never be as dominant as they once were, but as the computing market grows there’s a growing pie to share in regardless.
Basically, as a former Intel employee, Intel is ripe for being disrupted. (The employees only want to follow the rules and believe the rules will last forever.) Why it hasn't been disrupted yet, is a bit harder to explain.
Re:Intel let the chips fall where they might
by BuckRogers on Monday August 01, 2022 @02:00PM
(#32309788)
I'd argue that it's the holy grail of business, not the bible. I realize you weren't claiming it is.
Disruption is overrated. It's a lot harder to do than it is to think about or wish for. Neither Nvidia or AMD have "disrupted" Intel. Only Apple came close to that and even then only in theory, and only in their own walled garden market. If they were selling M-chips to everyone, you MIGHT have disruption, but that's easier said than done because there's an enormous list of tasks that need executed to a T that has to be completed. Everyone wants to be the disruptor, usually without the vision, talent, or execution, when "just making a lot of money" is also a fine goal that seems to be overlooked. Nvidia was always more likely to disrupt Intel with their ARM chips, but again, easier to dream about than execute. Hasn't happened. It's not that easy nor is it truly done often.
Especially the concept of sustainably making money is dismissed outright by nearly everyone. Here in the US it's all about the "bottom line" and quarterly executive bonuses. They love their reorgs. It's a broken model. While the Japanese, such as at Toyota, do a good job out of national honor. We can't compete with that sense of national pride. Western employees are broken culturally, and largely due to decades of watching the rich take it all as their prospects dwindle. Most of it lay at the feet of management though, as we made quality stuff in decades past.
I'd prefer to execute in a disciplined manner with quality output, something I could be proud of and fairly compensated for. Sustainably "just getting rich", Toyota style. That's not good enough in the US, we prefer booms and bust, then ending up broke. Best case for an American company is ending up a zombie corporation reliant on endless bailouts while claiming they're too big to fail. Even financial organizations that do nothing but sign documents as "product", reissuing Federal Reserve loans at usury rates, and would leave no gap upon disappearing overnight. To be replaced with a more rational actor that will be more responsible with such an easy way to make money.
At some point the Americans are going to have to realize that disruption can also be simply be delivering quality, reliably for decades. Toyota vs GM/Ford is a perfect example, as they were definitely disrupted. Being the market leader, dominating while getting rich is too longterm for us though.
Apologies for this excessive screed. This topic always bugs me. If I'm wrong, I'd appreciate it someone correct me so I can be a bit less pessimistic.
Re:Intel let the chips fall where they might
by klelatti on Monday August 01, 2022 @09:25AM
(#32305612)
The problem is that at the moment Intel only really have one business - selling CPUs for desktop/laptops and servers. In the latter market they are under severe pressure from AMD and Arm based CPUs. The former is relatively low growth.
They might want to have a third party foundry business but at the moment they don't have a credible one.
So they have massive historic and prospective capital investment and a cost base focused on difficult or uncertain markets.
Add to this poor recent execution and you have a toxic mix.
Re:Intel let the chips fall where they might
by pclmulqdq on Monday August 01, 2022 @09:52AM
(#32305942)
IMO at 14 nm (when it was the leading-edge process), they should have spun off the foundry business. They had accepted a few customers at 22 nm, which allowed them to work out the kinks as a pure play foundry, but instead they acquired their biggest customer. They wanted to keep their CPU hegemony so much that they couldn't see the big opportunity.
Re:Intel let the chips fall where they might
by klelatti on Monday August 01, 2022 @10:02AM
(#32306072)
Agreed. Also there has been a tendency to see losing / turning down the iPhone as the pivotal moment but I think it was the years after that which were crucial. They relentlessly pushed x86 on mobile when it was clear there was little or no appetite. If they'd built Arm based SoCs for Qualcomm etc (inc maybe Apple) then it might have been a very different story.
Re:Intel let the chips fall where they might
by klelatti on Monday August 01, 2022 @11:45AM
(#32307709)
My 'growth' comment referred to the laptop /desktop __market__ which medium / long term - which is what the investors will focus on - is clearly low growth.
Intel's single quarter performance in that market is quite a different thing altogether with short term post pandemic issues and Intel's own competitive struggles.
Re:Intel let the chips fall where they might
by pclmulqdq on Monday August 01, 2022 @09:55AM
(#32305968)
So far, I have been able to pull out a lot more perf/$ from Intel CPUs on AWS than from ARM. I know the ARM cores benchmark well for certain applications, but at the moment, the Intel SIMD instructions are a little too powerful, so if you can take advantage of them, you will win on x86.
Re:Intel let the chips fall where they might
by foobiekr on Monday August 01, 2022 @02:01PM
(#32309820)
It isn't like Apple hasn't jumped ship before. It remains to be seen if Apple can stay on top of the performance curve. Right now they are doing it, primarily, through capital intensive means. They have the best fab process, bought with capital investment and order guarantees, and some of the best ASIC and CPU arch heads in the world, bought with comp packages.
It's like like there's some law of nature that means Apple will always be on top. We'll see what Intel does - they might go down the RISCV path for all we know, wherein their own talent richness may well put them at an advantage.
Re:Intel let the chips fall where they might
by sliken on Monday August 01, 2022 @03:12PM
(#32310769)
Sure. But keep in mind that say Apple's future arm designs fail to be market
leading, without much pain they could switch to any number of other arm cores.
Similarly is TSMC falls behind there's other fabs out there, like Samsung, or even Intel.
Re:Intel let the chips fall where they might
by aNoob7000 on Monday August 01, 2022 @09:34AM
(#32305720)
I agree that Intel is in trouble on many front, but I don't see the x86 instruction set going away anytime soon. The desktop and server world is basically running on x86 at this time.
Re:Intel let the chips fall where they might
by dboreham on Monday August 01, 2022 @10:15AM
(#32306248)
There are big cracks appearing though in this edifice. Thanks to ARM Macbook, it's now expected that projects ship ARM packages, ARM containers etc. And therefor it's practical to deploy on ARM servers. Raspberry Pi also helped with this historically. Even Windows -- I use a Surface Pro X which provides a wonderful travel experience.
Re:Intel let the chips fall where they might
by sliken on Monday August 01, 2022 @02:15PM
(#32310054)
Things aren't going well. New fab shrinks and CPUs are years late. Apple ditched Intel chips for desktop/laptops. Various large clouds (like Amazon) are pushing ARM based systems. Even Microsoft is starting to push arm.
Promised fab shrinks and CPUs are years behind schedule. The sapphire rapids Xeon chip is in it's 12th revision and despite being 1.5 years late looks to be another 6-7 months out. It was originally going to compete with Zen3 Epyc chips and now looks like the Zen 4 Epycs will have a good head start.
Apparently the GPU effort is also years behind, failing to meet numerous previous timelines, and still has an uncertain future.
Optane never quite delivered and has been killed off.
So major setbacks in servers, fabs, desktop chips, storage and GPUs.
The semiconductor shortage has helped intel, better to buy an Intel server than wait a year for an AMD ... but that only works until AMD can meet demand.
Can Intel turn it around, sure. Will that happen? Seems far from sure. Today's situation has been years in the making and it's far from clear if the fixes have been made.
Re:Intel let the chips fall where they might
by colinmhayes on Monday August 01, 2022 @01:02PM
(#32309014)
Nvidia is worth like 3 times as much as intel. I believe it’s one of the 20 largest cap companies. Just doesn’t make sense to me when TSMC will suck most of their profit margins up in the medium run, especially if intel can’t figure it’s foundry business.
Re:Intel let the chips fall where they might
by freemint on Monday August 01, 2022 @08:10AM
(#32304860)
> The Holy Roman Empire was a pale shadow of the glory that was Rome. But still effective, mind you, give the times many centuries after the fall of Rome.
Sigh ... Don't use history analogies if you get them wrong.
Re:Intel let the chips fall where they might
by throwaway4good on Monday August 01, 2022 @08:42AM
(#32305143)
As it looks now we are heading to a semi slump with declining demand and oversupply while bailing out a disfunctional giant. My guess we are going to see fabs that are so challenged on revenue that they are unable to cover the operational costs. Let alone provide a return on the amount invested.
Re:Intel let the chips fall where they might
by azemetre on Monday August 01, 2022 @08:28AM
(#32305018)
There are very few companies that actually own fabs for mass production in the United States. AMD use to own fabs but they sold them quite some time ago.
By virtue of being one of the five companies (while also being the largest) you’re going to easily get the money.
As for why, fabs are a tremendous resource for a specialized economy and will easily become a national security risk due to how wide spread they are in the military and to sell advanced products.
There’s also a de globalization movement that’s been underway for nearly a decade. Having a country whose neighbors aren’t in a politically stable region of the world is a future disaster. It’s also proven to be a disastrous domestic policy where supply constraints crunches every company.
It’s likely going to take a decade to resolve the supply chain issues and it nearly takes half a decade to bring a new fab online.
Re:Intel let the chips fall where they might
by api on Monday August 01, 2022 @09:32AM
(#32305704)
China's entire industrial base is state subsidized. TSMC is a strategic asset of Taiwan. The US is still strongly dominant in aerospace, and there are few efforts or companies in that entire sector that are not subsidized (largely by defense spending and NASA). I could fill up several pages with this.
If nobody subsidized their industries, then maybe we could do just fine without subsidizing ours. In today's world, not subsidizing key industries is unilateral disarmament. It'd be like the USA unilaterally giving up all its nuclear weapons while Russia, China, North Korea, etc. still have theirs and hoping everyone plays nice.
Ask Russia how not having domestic chip fab capability is working out for them.
Re:Intel let the chips fall where they might
by formerkrogemp on Monday August 01, 2022 @10:10AM
(#32306169)
The efficient market rewards inefficient stock buybacks, record debt fueled dividends, years of underinvestment, low wages, and most of all no strings attached bailouts. They may be a loser, but we'll pay for their loss. Socialize the losses/costs and privatize the gains.
Re:Intel let the chips fall where they might
by pclmulqdq on Monday August 01, 2022 @09:56AM
(#32305987)
Settling for 1/10th those amounts years later (after the lawsuits look like they are going badly) would compensation for anticompetitive behavior. Settling early for the full price (or more) is a bailout.
Re:Intel let the chips fall where they might
by Retric on Monday August 01, 2022 @10:13AM
(#32306212)
Lawsuits aren’t limited to the exact damages, Intel settled early to reduce the risk of a massive loss in the future.
It’s important to realize the lawsuit could easily have survived past the point where AMD failed as a company. Just as patents get sold off so too can potential damages from lawsuits.
Re:Intel let the chips fall where they might
by pclmulqdq on Monday August 01, 2022 @11:23AM
(#32307325)
They explicitly told people that they settled early (and generously) to avoid antitrust risk. Intel could have bankrupted AMD with the process. They chose to pay them a lot of money instead. Why? A bailout.
Re:Intel let the chips fall where they might
by Retric on Monday August 01, 2022 @11:35AM
(#32307547)
They said that due to good marketing rather than admit wrongdoing, but being a monopoly isn’t illegal. In fact they likely qualified as a monopoly at the time.
Re:Intel let the chips fall where they might
by pclmulqdq on Monday August 01, 2022 @09:48AM
(#32305898)
I think that given AMD's balance sheet at the time, and the state of the lawsuits, they would have allowed them to run their course for a lot longer. Giving your competitors generous settlements at the beginning phases of a lawsuit is not usually good practice.
Re:Intel let the chips fall where they might
by sliken on Monday August 01, 2022 @03:32PM
(#32311010)
Because is TSMC was unavailable to the USA it would take many years for the USA to recover. This bail out gives the USA a head start on a TSMC replacement so that we don't have to halt production of GPUs, CPUs, phones, tablets, appliances, cars, trucks, cranes, tanks, missiles, etc. etc. etc.
Re:Intel let the chips fall where they might
by ksec on Monday August 01, 2022 @08:11AM
(#32304869)
Yes, but none of them are competitive. With Modern CPU you need Memory with ultra high bandwidth, which no persistent memory offers. With low latency storage it doesn't compete against Z-NAND or SLC NAND with special tuning.
Re:Intel let the chips fall where they might
by ksec on Monday August 01, 2022 @10:01AM
(#32306050)
Not AFAIK, commercially speaking. At least not in the next 5 to 7 years. And I would love to be wrong.
DRAM and NAND are so well optimised in terms of cost and scale ( Which was the reason why Optane failed to compete ). It is like the lithium Battery. Whenever there is a "Revolution Battery", it has to pass the check list [1].
Re:Intel let the chips fall where they might
by georgeburdell on Monday August 01, 2022 @10:25AM
(#32306395)
Many foundries, including Intel and TSMC, offer MRAM through their foundries. As far as I know, no plans for a customer to use it in the Optane space though
Re:Intel let the chips fall where they might
by galogon on Monday August 01, 2022 @08:40AM
(#32305125)
I still don't understand how Apple, a lifestyle company, managed to beat Intel at their own game. It must be humiliating for the employees and the execs. I wonder what the morale is inside the company.
Re:Intel let the chips fall where they might
by kmeisthax on Monday August 01, 2022 @09:34AM
(#32305722)
Apple isn't a lifestyle company. They're just a tech company that happens to use lifestyle marketing as their selling point. But even before they owned a chip design firm their key strength wasn't the marketing; it was superior UX design.
Or to paraphrase Papa Steve: the only problem with other tech companies is that they have no taste.
Related note: Intel was in the bidding process for the original iPhone chip and lost to Samsung, partially because Intel's chip was too expensive for their liking. Apple really expects its vendors to be able to take a haircut on unit pricing and make it back up in volume, which isn't exactly Intel's thing. Had Intel met Apple's requirements they may have dissuaded Apple from getting into chip design in the first place.
Re:Intel let the chips fall where they might
by alwillis on Monday August 01, 2022 @09:57AM
(#32306002)
It’s this type of thinking is why Intel is in the mess they’re in.
Calling Apple a lifestyle company kind of diminishes the relentless innovation they’ve demonstrated for decades.
It reminds me how Microsoft, Palm, Nokia, Blackberry dismissed the iPhone in 2007. That didn’t turn out so good for them.
I’m old enough to remember when Intel was an ARM licensee… I’m sure there are some Intel engineers lamenting how all of this could have been different.
Re:Intel let the chips fall where they might
by r00fus on Monday August 01, 2022 @12:04PM
(#32308059)
Comparing a market for devices to all of computer hardware (including peripherals/components) is what's ridiculous. The more apt analogy is computers (PCs) to iPhones.
Re:Intel let the chips fall where they might
by chmod775 on Monday August 01, 2022 @09:03PM
(#32314076)
Throwaway4good was using the comparison of iPhones to PCs to allude that Apple has more money available to develop silicon, yet the pre-built PC market is a tiny subset of where money is made using semiconductors.
That's why it's a misleading claim.
I didn't expect I would have to elaborate this beyond what I said. My bad.
Re:Intel let the chips fall where they might
by ckastner on Monday August 01, 2022 @07:05AM
(#32304317)
Tangentially: The GDPR has been around for 4+ years now. They should know that a cookie consent popup where you have opt-out for 15 processing purposes individually almost certainly won't fly, and having to opt-out of 100+ vendors individually definitely won't fly.
Re:Intel let the chips fall where they might
by MomoXenosaga on Monday August 01, 2022 @11:56AM
(#32307909)
Until EU starts issuing international arrest warrants (don't hold your breath) Americans will do whatever they want.
Try to avoid doing business with US corporations.
AMD and Nvidia will always be beholden to TSMC - yes, they avoid the capital outlay and risk of foundries, but they also give up the possible upside of owning the foundry.
Perhaps intel will never be as dominant as they once were, but as the computing market grows there’s a growing pie to share in regardless.
Basically, as a former Intel employee, Intel is ripe for being disrupted. (The employees only want to follow the rules and believe the rules will last forever.) Why it hasn't been disrupted yet, is a bit harder to explain.
Disruption is overrated. It's a lot harder to do than it is to think about or wish for. Neither Nvidia or AMD have "disrupted" Intel. Only Apple came close to that and even then only in theory, and only in their own walled garden market. If they were selling M-chips to everyone, you MIGHT have disruption, but that's easier said than done because there's an enormous list of tasks that need executed to a T that has to be completed. Everyone wants to be the disruptor, usually without the vision, talent, or execution, when "just making a lot of money" is also a fine goal that seems to be overlooked. Nvidia was always more likely to disrupt Intel with their ARM chips, but again, easier to dream about than execute. Hasn't happened. It's not that easy nor is it truly done often.
Especially the concept of sustainably making money is dismissed outright by nearly everyone. Here in the US it's all about the "bottom line" and quarterly executive bonuses. They love their reorgs. It's a broken model. While the Japanese, such as at Toyota, do a good job out of national honor. We can't compete with that sense of national pride. Western employees are broken culturally, and largely due to decades of watching the rich take it all as their prospects dwindle. Most of it lay at the feet of management though, as we made quality stuff in decades past.
I'd prefer to execute in a disciplined manner with quality output, something I could be proud of and fairly compensated for. Sustainably "just getting rich", Toyota style. That's not good enough in the US, we prefer booms and bust, then ending up broke. Best case for an American company is ending up a zombie corporation reliant on endless bailouts while claiming they're too big to fail. Even financial organizations that do nothing but sign documents as "product", reissuing Federal Reserve loans at usury rates, and would leave no gap upon disappearing overnight. To be replaced with a more rational actor that will be more responsible with such an easy way to make money.
At some point the Americans are going to have to realize that disruption can also be simply be delivering quality, reliably for decades. Toyota vs GM/Ford is a perfect example, as they were definitely disrupted. Being the market leader, dominating while getting rich is too longterm for us though.
Apologies for this excessive screed. This topic always bugs me. If I'm wrong, I'd appreciate it someone correct me so I can be a bit less pessimistic.
They might want to have a third party foundry business but at the moment they don't have a credible one.
So they have massive historic and prospective capital investment and a cost base focused on difficult or uncertain markets.
Add to this poor recent execution and you have a toxic mix.
---
1: https://www.intc.com/news-events/press-releases/detail/1563/...
Intel's single quarter performance in that market is quite a different thing altogether with short term post pandemic issues and Intel's own competitive struggles.
It's like like there's some law of nature that means Apple will always be on top. We'll see what Intel does - they might go down the RISCV path for all we know, wherein their own talent richness may well put them at an advantage.
Similarly is TSMC falls behind there's other fabs out there, like Samsung, or even Intel.
Promised fab shrinks and CPUs are years behind schedule. The sapphire rapids Xeon chip is in it's 12th revision and despite being 1.5 years late looks to be another 6-7 months out. It was originally going to compete with Zen3 Epyc chips and now looks like the Zen 4 Epycs will have a good head start.
Apparently the GPU effort is also years behind, failing to meet numerous previous timelines, and still has an uncertain future.
Optane never quite delivered and has been killed off.
So major setbacks in servers, fabs, desktop chips, storage and GPUs.
The semiconductor shortage has helped intel, better to buy an Intel server than wait a year for an AMD ... but that only works until AMD can meet demand.
Can Intel turn it around, sure. Will that happen? Seems far from sure. Today's situation has been years in the making and it's far from clear if the fixes have been made.
I don't write Intel of but they have a LOT of catching up to do.
Sigh ... Don't use history analogies if you get them wrong.
By virtue of being one of the five companies (while also being the largest) you’re going to easily get the money.
As for why, fabs are a tremendous resource for a specialized economy and will easily become a national security risk due to how wide spread they are in the military and to sell advanced products.
There’s also a de globalization movement that’s been underway for nearly a decade. Having a country whose neighbors aren’t in a politically stable region of the world is a future disaster. It’s also proven to be a disastrous domestic policy where supply constraints crunches every company.
It’s likely going to take a decade to resolve the supply chain issues and it nearly takes half a decade to bring a new fab online.
edit: spelling
If nobody subsidized their industries, then maybe we could do just fine without subsidizing ours. In today's world, not subsidizing key industries is unilateral disarmament. It'd be like the USA unilaterally giving up all its nuclear weapons while Russia, China, North Korea, etc. still have theirs and hoping everyone plays nice.
Ask Russia how not having domestic chip fab capability is working out for them.
Say force them to open source chips older than a decade.
I wouldn't want us to fall behind on chip tech though.
AMD was bailed out for years as Intel was king, and it’s a good thing they were.
It’s important to realize the lawsuit could easily have survived past the point where AMD failed as a company. Just as patents get sold off so too can potential damages from lawsuits.
You think Intel would settle all those lawsuits if Intel did nothing wrong?
Not the same thing at all.
We have to pay to play. Unless we're cool with not manufacturing chips domestically.
Not to mention most bailouts don't have clawbacks like the CHIPS act.
DRAM and NAND are so well optimised in terms of cost and scale ( Which was the reason why Optane failed to compete ). It is like the lithium Battery. Whenever there is a "Revolution Battery", it has to pass the check list [1].
[1]https://news.ycombinator.com/item?id=26354453
Or to paraphrase Papa Steve: the only problem with other tech companies is that they have no taste.
Related note: Intel was in the bidding process for the original iPhone chip and lost to Samsung, partially because Intel's chip was too expensive for their liking. Apple really expects its vendors to be able to take a haircut on unit pricing and make it back up in volume, which isn't exactly Intel's thing. Had Intel met Apple's requirements they may have dissuaded Apple from getting into chip design in the first place.
Calling Apple a lifestyle company kind of diminishes the relentless innovation they’ve demonstrated for decades.
It reminds me how Microsoft, Palm, Nokia, Blackberry dismissed the iPhone in 2007. That didn’t turn out so good for them.
I’m old enough to remember when Intel was an ARM licensee… I’m sure there are some Intel engineers lamenting how all of this could have been different.
Can we call this "the Nokia effect"?
If you meant to refer to the entire computer hardware market, you're wrong by an order of magnitude.
iPhone Revenue 2021: $191.9 billion
Computer Hardware Revenue 2021: $1,129.39 billion
That's why it's a misleading claim.
I didn't expect I would have to elaborate this beyond what I said. My bad.
Even Google added a "Reject all" button, eventually.
Don't worry Intel, Google and their "strategic AI assets" are only a few years behind you in line, you'll soon have company
It's fitting given the name, no?