I'd be more interested in hearing about the terms of the Qualcomm deal. Is Qualcomm putting its eggs in the Intel basket for 2025 or has Intel reached an agreement with Qualcomm that Qualcomm will buy from them if it fits Qualcomm's needs and it's ready in time?
If the agreement is "Qualcomm will send some of its business Intel's way," that isn't a huge vote of confidence. Heck, are we talking about Qualcomm's Snapdragon 8xx series or just some Qualcomm chips like the ones they put in WiFi routers? Or their low-end Snapdragon 4xx series?
The article is kinda spinning it as "Intel will be making the Qualcomm chips going into everyone's phones," but it could be more "Qualcomm is going to have Intel manufacture its low-end stuff which is always behind tech-wise and Intel is giving them a big discount because Intel needs a win."
> In the chip world where smaller is better, Intel previously used names that alluded to the size of features in "nanometers". But over time the names used by chipmakers became arbitrary marking terms... This, he said, gave the mistaken impression that Intel was less competitive.
They are definitely less competitive at this point. Nanometers might not be the right measurement, but one can measure transistor density. Maybe that would make more sense?
Even if "Intel 7" (10nm Enhanced SuperFin) is equivalent to TSMC's 7nm, we saw TSMC's 7nm back in 2018. I have a TSMC 5nm in my pocket. I guess if Intel can keep to its road map, it will catch up. If "Intel 4" comes out with products in early 2023, it won't be far behind TSMC's 3nm process (which will likely debut in the 2022 iPhone).
Of course, we'll have to see how well these Intel chips perform. We should be able to benchmark Intel 7 processors with Alder Lake processors landing later this year. Will these stand up well against AMD's 7nm Zen 2/3? Will they stand up against AMD's upcoming 5nm Zen 4? I guess we'll be able to test Intel's chips against what AMD has been shipping and see how well Intel 7 stands up against both 7nm and 5nm processes.
Disclaimer: Qualcomm employee with no inside knowledge of the terms of the deal, opinions are my own.
Generally fabless semiconductor companies use a variety of foundries and processes for their products. It largely depends on the technical requirements of the particular project as well as the cost of the node. In the early stages of a product development cycle, the team will choose a foundry/process and stick with it to the end. In rare cases the process will switch mid-stream, but this is costly because design essentially has to start over. (For digital, this would be after RTL is frozen; for analog/RF, a change at any point requires starting over)
Sometimes, second sources are added after the first source is in production. These tend to be the lower cost fabs like SMIC or UMC, but they in turn design their processes to be drop in replacements for top fabs like TSMC and Samsung to make it easy to switch over.
Given the above, and that Intel's process is not aimed at copying TSMC/SEC, I would guess that you'll slowly see a handful of non-critical products rolling out on "Intel 7". If the process turns out to be performant enough, you'll see more, even possibly higher end more critical product lines. Don't forget that Qualcomm isn't just about Snapdragon. There is a boatload of RF, Power Management, IoT and WiFi as well.
It doesn't necessarily require a massive design change. The problem is that nearly every aspect of two different processes is different, even when there is supposed to be an "easy" migration path, such as when the new process is just an optical shrink of another process.
Specifically, the transistor properties are different, and so are the properties of the tiny metal wires connecting the transistors together. Once you get to the design phase where you're dealing with individual transistors (for analog this is right off the bat; for digital it's once you get to place & route), you spend thousands of hours simulating the behavior of the analog circuits, or the effects of the particular arrangement of digital logic gates, to make sure everything works. When the transistor and wire properties change, you at minimum have to repeat all of that verification
Sometimes, the new process is different enough that an old design no longer works, and it takes a lot of engineering effort to fix it. Other times, a few small tweaks are enough. In some cases, particularly for digital-only ICs, the foundry will validate the migration between two processes, essentially saying, "we guarantee the IC you already designed will just work in this new process". Smaller companies might just go with it and accept any yield fallout, but bigger companies like Qualcomm would probably do their own independent verification.
so if I can summarize -- what we think of as a logical transistor starts to have all sorts of physical properties, some dictated by the material, the physical layout, and probably what's near it. Designers spend piles of time simulating this, and a process change requires that work to be redone.
You got it right. Translating from a logical design to a physical design requires following rules laid out by a foundry for a given process node (5nm, 7nm etc.). To check if a physical layout confirms to a node, Design Rule Checking [1] is performed as part of the Physical Verification step in EDA [2]. The simulation/checking is highly complex and takes hours (sometimes days) even with sufficient parallelization. Once a design passes the checks, it is deemed ready for manufacture by the foundry.
I have been doing semiconductor digital physical design for over 20 years
The Verilog RTL for the digital portions of the design are often identical for the same design in 2 different foundry processes.
The combinational logic gates (AND/OR, registers, etc) are made out of transistors and the transistors are different so all of the logic cells are different.
The foundry gives you a library with these cells. You use a synthesis program to "compile" them from Verilog to logic gates.
One foundry may have 16 different types of AND gates of various sizes, drive strength, and power usage. The other foundry may have 14 types but their 14X cell is stronger than the 16X cell of the other foundry. There is no direct mapping between these.
One foundry may make all standard cells 200 nm in height. They are then placed in uniform rows by the automatic placer tool (Cadence Innovus or Synopsys IC Compiler)
The other foundry made their cells 220 nm in height. The block sizes in your floorplan are now going to be different.
One foundry made their process for their internal mobile phone use so they focused on small cells for low power. Now you try to make your $3,000 GPU in this process and it won't work as well.
One foundry made metal layers 1-5 thin and 6-10 thicker and 11-15 really big. The other foundry had metal layers 1-6 thin. In your old chip you decided to use metal 6 for a certain power layer. Now it is thin and it isn't good for carrying power but thin wire is better for signals. More stuff to change.
It gets even worse for custom analog stuff like a PCIE/USB Serdes or PLL. The rules are even more different. You may need some high voltage area and a larger spacing between that section but the other foundry let you put it closer together.
In my experience TSMC is the easiest to work with. They have a huge infrastructure to help external customers.
Samsung foundries are mainly used by Samsung so they aren't great to deal with as an external customer.
Intel only has external customers for strategic things like Altera which they bought 2 years later. Intel optimizes their DRC rules to make one chip a billion times. Their DRC rules are very restrictive to optimize yield. This takes more people and time.
TSMC is optimized for a thousand customers making a thousand different chips. They don't care if your yield is 1% lower and you probably don't either if you can get to market 3 months earlier with 10% fewer employees.
For Intel that extra 1% over a billion chips is worth hiring more people to meet those extra DRC rules.
Given that almost all previous Intel foundry customers have failed, why should we expect a different result this time? Achronix is perhaps the most telling as FPGAs are incredibly regular and should have been a natural fit for Intel's manufacturing capabilities, but even Achronix ended up moving to TSMC.
Right now Altera FPGAs seem to have the same supply constraint issues as Xilinx at present (my favorite check is to look at stock levels on Digi-Key), so I'm getting the impression that Intel is unable to crank them out any better than other foundries. Furthermore, if Intel really was determined, I'd have expected more than just a handful of Altera's SKUs to be on Intel's process 5.5 years after acquisition.
My faith in Intel is not strong on this front. Does anyone else have reason to believe this time will be different?
thank you -- I built a (tiny) processor for class with Mentor Graphics, but laying out logical gates was as far down as I went. This is really fascinating and thanks again for sharing.
There was a company called Allwinner Semiconductors. It was making super cookiecutter SoCs for tablets, and been aiming at the exploding "phablet" space.
In the tablet space, it completely vanquished companies with many times the market cap, and marketing expense Qualcomm included.
Qualcomm was really scared of them following them to the smartphone space, and going for their jugular.
What they did:
In around 2015-2016, they announce a "strategic deal" with Allwinner, with Allwinner distributing their low end SoCs on a borderline giveaway terms.
Just few days later, their "phablet" offering disappears from their website. Few month later, their C-levels leave the company one after another, with some leaving for the fancy life in the US.
I had contacts of some of their management people in my wechat account, and after 2016 their feeds were all cheese, and wine, cheese, and wine, and fancy dining, and parties.
Nowadays, Allwinner is a shadow of its former self in its heydays, mostly dealing with most basic settop box SoCs.
I miss the days of weird phones and tablets with brand names I've never heard of with Allwinner SoCs in them. Such odd devices, so much fun to play with and explore. Rockchip based handheld gaming devices are about the closest to it nowadays.
Yes, but that's not what the poster is talking about. The Pinephone is a comparatively conventional/mundane handset design. It sounds like they were longing for more 'out there'/experimental designs that apparently were using A64 several years ago.
> but one can measure transistor density. Maybe that would make more sense
Though by stacking 2D layers they can artificially balloon their density figures: 7nm is nominally ~95 million transistors per mm2 - but stack two of them and you can boast having 185MTr/mm2
Different number have different amounts of meaning. The technology is so incredibly complicated at this point that coming up with a single number that can be used for meaningful comparisons doesn't appear to be possible.
The nanometer measurement was used to measure the whole transistor and when it was becoming harder to go smaller the nanometer measurement was changed to the smallest feature you can make on a silicon, so the competition is now who can make the best transistor.
Not sure why you're being downvoted, you're absolutely correct.
The low-end 400 series is a good 1-2 processes behind, being 11nm currently with the Snapdragon 480 5G (2021) on what looks to be Samsung's 8nm (aka, Samsung 10nm+), and the Qualcomm 215 (2019) is on 28nm even.
Qualcomm isn't going to buy bleeding-edge foundry space for a low-end product, and definitely not for a wifi router which has almost no power or significant performance constraints.
My (limited) understanding is that a lot of beamforming / MIMO techniques are limited by the power budget of the WiFi chips. So if you could use the fanciest processes, the WiFi chips could be better. I don't know if that's worth the price (everyone expects WiFi to suck 24/7), or if the process helps there, but it does sound like manufacturing technology, not math, is the limiting factor for WiFi right now.
Intel 7nm should be equivalent to TSMC 5nm and should be launching some time soon (if not already). It didn't seem to bite off too much new stuff at one time like 10nm did, so I doubt it will see the same delays.
Intel's fab woes are starting to lessen. I have little doubt that they'll wind up leading the pack again in the next couple years. I hope that this time around there are more companies that can use those advanced fabs.
I wouldn't go so far as to say "buy Intel stock" because institutional investors have likely already adjusted to the announcement and it's likely priced in.
That said, I think government encouragement is a critical piece of the puzzle and should derisk this operation. The recent supply shocks to the computer industry, when component prices and lead times shot up independent of any geopolitical conflicts (other than COVID), must have been a wake-up call to many that the US's relative lack of foundry capabilities has real strategic implications.
Adding more money to a crappy organization tends to make it crappier. I don't see evidence that money availability is Intel's problem, but it does appear that motivation/near-term thinking is a substantial problem.
Oddly, this is a pretty similar pattern to what we've seen with other companies before they divest/outsource manufacturing. Their internal efforts stumble due to lack of drive/funding, followed by a lucrative buyout or sale of their manufacturing operations to an oversees buyer.
I don't understand why executives of American companies don't see the manufacturing/production side of their companies as anything more than a cost center.
> I don't understand why executives of American companies don't see the manufacturing/production side of their companies as anything more than a cost center.
This is interesting to me, because I think of it as the stereotypical (if not always accurate) description of American business. The common criticism of American companies is that they outsource productive capacity, go "asset-lean", lever up on their intellectual property, over-financialise etc etc. I've generally thought of the preference for investing in "manufacturing/production" to be a populist sentiment, and out of favour in the boardroom.
There are certainly plenty of examples of both types, but in the semiconductor space the trend is clear -- manufacturing has been outsourced from the US, and the old big chip brands mostly stopped making their own chips.
And until recently I've been mostly sympathetic to (what I think is) your view, which I've also ascribed to the stereotype of "corporate America" -- that companies should put their capital to its most productive work, and that mostly that means "intellectual property and innovation" or some similar vague notion. Ricardian comparative advantage and all that.
One thing that's given me pause has been arguments like this: https://berthub.eu/articles/posts/how-tech-loses-out/. Very roughly, that production involves a necessary kind of innovation, and that intellectual property advantages can be less durable than process advantages.
If you haven't read "Outsourced Profits" on how and why this process occurred at Boeing and how it ended up causing problems (referenced in the above article) I'd say the document is well worth your time.
It's actually not priced in. Price is almost as low as it was before Pat came on as CEO. Subsidies from the US ($50B for the industry) and EU ($20B likely, not confirmed yet) were announced months later.
Intel's selling at something like 12x earnings. The fab business is another devision in the company, not an alternative to selling chips.
With their yearly cadence, if they deliver, they will take back the share they've lost and then some.
Their designs are still superior to AMD, they have just been slower to finish them and roll them out. With improved execution and vertical integration, they should be beating competitors on performance, efficiency and price.
If they stick to their new product road map, they should be taking back market share from AMD in full by 2023.
It's an existential question. Right now the US economy is entirely dependent upon Taiwanese semiconductor fabrication. This isn't a glaring problem yet, because China is also dependent upon the same supply chain, giving rise to a mutual incentive not to rock the boat too hard. However, China has dedicated substantial resources to building its own fab infrastructure since well before the pandemic [0], with the goal of establishing a parallel ecosystem that circumvents the US-controlled supply chain [1]. If the US does not follow suit, its dependence upon Taiwan becomes a massive liability.
How much did we spend on the Iraq War, despite our own country being a major producer of oil and gas? Our economic incentive to entangle ourselves in Taiwan is even greater, but the stakes of military confrontation with China would be more dangerous. How much would you pay to avert such a conflict -- or worse, to avert the consequences of acquiescing to China's expansionism? Ironically, dependency on Taiwanese semiconductor fabrication reduces our ability to defend the island. Until the time that a fab interruption there would not halt our economy, Taiwan remains a US pressure point and China will continue to press it.
I think $200 billion is a cheap price tag for a path forward that involves neither warfare nor perpetual paranoia about every perturbation in the South China Sea. Taiwan established itself as a semiconductor powerhouse largely because its government backed TSMC. South Korea's government bends over backwards for Samsung, which makes up about a fifth of their GDP. The way forward for the US may be similar in mentality: throw the government behind the industry. US fabs need resilient and high-quality infrastructure for power, water, and logistics. We can use the collimated power and authority of the government to yield a vastly more effective long-term solution than piecemeal tax credits on lithography scanners.
> Right now the US economy is entirely dependent upon Taiwanese semiconductor fabrication.
I'd be curious to see this quantified in some way. TSMC has a lead at the moment for bleeding edge fabrication, but Intel isnt that far behind and is massive in scale (revenue is nearly double TSMC, though its not an apples to apples comparison). A lot of the most impactful shortages this past year were in non-sexy parts like microcontrollers and other lower end chips that can be built in many different countries, since they are fairly simple (28nm, 40nm and even larger processes are commonly used).
The world is intertangled enough that losing Taiwan for some reason would be pretty bad, but I cant think of another country better equipped to handle semiconductor production than the US (Korea and Japan do well also, but at smaller scale than the US).
Their #2 company has failed miserably and exposed as fraud. SMIC is the only game in town in China.
>> How much did we spend on the Iraq War, despite our own country being a major producer of oil and gas?
With all due respect, from 2004 - 2007, all we were talking about is Peak Oil and US was not such a major producer then as we are today. So some recency bias here.
>>The way forward for the US may be similar in mentality: throw the government behind the industry.
Yes, US will become mercantilist again, and I will rejoice but the goddamn world will burn or not.
FYI: the automotive chips causing the crisis were made by two factories in Japan that caught fire (Renesas was one), not Taiwan. Needless to say, critical dependencies need to be onshored.
automakers is the number one reason for chip shortage. they believe car sales will drop during covid peak but they were wrong. they didn't order from their suppliers thus suppliers didn't sign wafer agreement with foundries. i believed we going to have a over supply issue in the future with all these fabs being build and the demand will wane overtime
I'd be surprised if the autos are much beyond low single digits (if that) in terms of share as global consumers of semiconductors. The global auto market produces a combined ~70 million vehicles annually vs >250 million PCs, >1.25 billion smartphones etc. with just about everything these days containing at least a microcontroller.
Sure, the semi shortage is a big deal to the auto industry, but the autos are almost rounding error to the semi industry.
There are about 100 MCUs per car. Even rear-view & side-view mirrors have their own micro-controllers. I believe the most common use are for voltage regulation.
Sure but quantity doesn't tell the whole story. Most of them are relatively tiny, in terms of die size. (obviously not so in the case of the more advanced end of the market such as Tesla, but rather the more common vehicles most are currently driving) So the point still stands that automakers aren't an out-sized user of silicon as the parent claims.
> Right now the US economy is entirely dependent upon Taiwanese semiconductor fabrication.
Strong disagree. What kind of application would not be able to run at all on Intel's latest processors? I can't think of anything aside from maybe some future development of VR.
Most HNers deal with software engineering, but the rest of the world simply does not have to have the latest and greatest. Assuming Intel stays in the back and the West has to use only Intel, applications will be a bit slower, processors will use a bit more power, there will be more focus on software efficiency and parallelism, and more cloud streaming. Hardly "entirely dependent".
This is dangerously careless thinking which completely disregards the complexity of supply chains.
Modern IT infrastructure needs lots of different kinds of chips, not just CPUs. Most of them aren't manufactured in the US. If chips stopped shipping from Taiwan tomorrow, then sure, people would scramble to switch to other manufacturing, and presumably some of the gap could be filled from existing products or of Korea, but getting a design up on a new process can easily take a year, often more. Not to mention that building up raw fab capacity takes time as well.
The impact on supply chains in the short term would be devastating even if the long term can be gotten under control. If the short term includes a war with China, then, well... it certainly complicates any sort of military scenario.
>but getting a design up on a new process can easily take a year, often more.
That's only relevant to companies that have to build all their own CPUs.
The world is not HackerNews. 99.999999% of companies can do with the radical idea of using an existing older, slower, perhaps more power hungry processor. [It can be argued this already happened - all the Spectre/Meltdown mitigations costing performance, and the world kept trucking on]. Apple's bottom line may be temporarily hurt, and AMD will be in trouble, but that's hardly an economy-shaking threat.
>Not to mention that building up raw fab capacity takes time as well.
TSMC is actually rather small in the raw picture.
>The impact on supply chains in the short term would be devastating even if the long term can be gotten under control.
We were just having a huge pandemic. We can handle merely another chip shortage.
>If the short term includes a war with China, then, well...
> Our economic incentive to entangle ourselves in Taiwan is even greater, but the stakes of military confrontation with China would be more dangerous.
> How much would you pay to avert such a conflict -- or worse, to avert the consequences of acquiescing to China's expansionism?
These are questions easier to answer than what most people think.
There are countless unclassified defence studies with very good estimates of how much would USA loose in case of an all out war with a nuclear state for worst, medium, best scenarios. I'll encourage readers to check them themselves.
Washington defence planners are most well aware of these things.
The later question is much more important, and with more real world relevance.
> consequences of acquiescing to China's expansionism?
There is no limit on on how much "acquiescing" you can do before such states will come up with more claims.
Remember "Finlandisation" precluding the WW2, and how it ended.
Giving the most minimal concessions to such opponent will instantly lead to them wanting more, and more, until you will have to retaliate, but now against much stronger enemy.
There is no "strategy," no political trick that works against opponent planning to attack you from the start no matter what. All treaties, political manoeuvres, compromise moves, even overt tribute payments will be only delaying the inevitable encounter in cul de sac.
In Russia, people gave a simple way to characterise behaviour of such people: "A guy who wouldn't stop until punched in the face."
Finlandization started very much before the WW2, as major power were making compromises with regional hegemons, acceding to their claims on their neighbours "as something natural"
It's big power saying "let them have it" basically, and to disastrous results every time.
Yes that happened, but that's not called Finlandization. I know this is pedantry now but the term is pretty specific and not applicable to countries of roughly equal power.
I believe the US military would easily be willing to plow $10B.
Now, does that mean intel will actually be effective? I have my doubts, baring some sort of shotgun approach where multiple experimental fabs are done simultaneously.
I have been ragging Intel for a long time since the NetBurst debacle, but it's still a massive company with the x86 cash cow for at least another half-decade until AMD closes a bit of a gap.
They have the resources and history to recover. I have heard references to massive managerial layer problems and treating contractors like crap and the brain drain of the old guard, but those aren't intractable problems.
Intel has responded in the past. AMD looked completely dead in the Hector Ruiz waning days, and look at them now. Empower some good engineers and watch what happens.
Intel needs to start swinging the money around then because their reputation is that they underpay by paying 'market rate' ie as little as they can get away with, and not by trying to be market leaders. Their managerial layers have destroyed their reputation and its going to take a lot to steer that into being something people say 'I want to work at intel' when given a choice of nvidia.
Underpaying engineering in an engineering company.
Almost guarantee what that is: management building a huge pyramid reporting structure and then enriching themselves. To avoid reducing profits (which would threaten the bigwig C-level strike price options), they simply steal from the "proles", in this case the engineers that made the company.
Good news is: it's easy to clean house on shitty managers. Just drop half the layers or more, and then hire or acquire some good.
The other thing Intel had going for it back in the day: there were at least a couple teams working on x86 execution, so when Netburst crapped out they at least had the Israeli team working on the mobile architecture to switch to.
Intel's big enough that if you need two microarchitecture approaches, or even need to start an ARM project, well, they have the revenue if they want the talent.
I mean, look at what Apple did, they practically built the top-end chip in mobile within, what, four or five years? I guess I don't know exactly when they started working on an in-house A-series processors vs when they took leadership. But it didn't take long it seems, certainly relative to the four-year wishywashy timeframe being suggested now.
I'm not sure how much difference that would make. Intel already has $20B in cash reserves and is still making large profits (in the billions of dollars). Their problems have nothing to do with lack money.
Case in point, the previous administration had already allocated $15 billion to a wall by the time they left office. A smaller amount that I'm sure both parties would agree is good for both national security and the economy should have no issues getting funded.
it's very often the case that US companies sacrifice national interests to save their insane profit margin -- Apple being one most notable example (or Nike who recently declared "the brand was of China and for China." (duck)
I'm pretty sure Intel will eventually triumph and reclaim their crown, but I wonder if what more the gov't can do to speed up their way to the top? I don't think the company was held back by red tapes or lack of R&D, capital funding (or tax incentive).
Just because the U.S. government gives Intel money doesn't mean it's a good deal for shareholders. I can think of two different ways this can go poorly:
1. It causes Intel to optimize for government money, similar to the many aerospace contractors
2. The government could acquire a stake in Intel, as it did with many companies under TARP during the 2008 Financial Crisis
Personally, I really hope we keep all 3 of these companies (and maybe gain a 4th competitor in the form of SMIC) for as long as possible. Having a variety of companies producing chips is important both for competition and supply chain resilience reasons.
> It operates five manufacturing sites in Burlington, Vermont, East Fishkill, New York, Malta, New York, as well as in Dresden, Germany and Singapore, allowing the company to serve a growing portfolio of more than 250 global customers, including many of the world’s largest semiconductor companies.
That's a lot of USA-locations. I'm not sure what your point is exactly?
>> (iirc Samsung is currently making chips in Texas actually, but that's relatively recent)
Samsung has been making chips in Austin, TX since late 1990's and by mid-90's the company was making most of Apple's APs there -- remember when Tim Apple defended its China outsourcing practices by pointing out that "the engines" of their portable devices were in fact made in the US in a rare CBS interview in early 2010? Apple quickly switched to TSMC in Taiwan after that.
It seems like Samsung in turn turned their Austin production to mostly memory production lines (and some logic chips for Tesla) after Apple left for China/Taiwan.
People who care about 'American chips' aren't distrustful of Taiwan or South Korea. In fact, they are really good allies to us.
What we are worried about is China's increasingly large Navy, their raw ambition, and the weakening of the US Navy (due to budget cuts planned out decades ago).
China may decide to attack Taiwan, which would cause Taiwan to lose chip production, which would hamper the US.
TSMC making a backup fab in the USA would be beneficial to both Taiwan and the USA. If Taiwan gets attacked, they can still supply chips.
It's about physical security afforded by the US mainland. It's a lot harder for China to attack Arizona than it is to attack Taiwan.
> the weakening of the US Navy (due to budget cuts planned out decades ago
Relative to the height of the Cold war, sure, but the US Navy still has 5 times more ( and much bigger) aircraft carriers than the next navy; and leads the other navies with similar multiples in submarines, various types of ships, airplanes, rocket and anti-rocket capabilities etc. etc.
The US Navy has a decent-ish chance in an engagement with the Chinese Navy for at least the next few decades, assuming the engagement takes place outside of the range of the Chinese air force.
> The US Navy has a decent-ish chance in an engagement with the Chinese Navy for at least the next few decades, assuming the engagement takes place outside of the range of the Chinese air force.
The issue with Taiwan is that any realistic engagement of China vs Taiwan means that Chinese Navy is in 100% support of its Air Force.
You're right. US Navy still dominates China outside of its missile / air force range. But China likely sees an opportunity to make a move vs Taiwan within the next 10 years (before this US Navy issue is patched up with the next generation of warships).
The next stage for China vs US potential combat isn't Madagascar... its Taiwan. A nation that's well within range of China's cruise missiles, air force, and mainland (so easily resupplying Chinese Navy). The US would be fighting with significant disadvantages if a Taiwan war were to ever break out.
I wouldn't invest anytime soon. Intel is so focused on making every single foundry workflow identical that they lose the flexibility that is more and more sought after these days.
It is kind of crazy how Intel's strength has turned into such a weakness. Qualcomm will eventually figure out that Intel is way too inflexible execute on their business.
Stick to mass production Intel, it's what you are OK at.
The fact that they've replaced their CEO with a highly-respected (within the company) former engineer and they've been rehiring a lot of other veteran former employees who were involved in many of their historical successes is certainly a good sign.
It's still a tall order, but my reading is that they seem to be doing all the right things. And they certainly aren't short of cash to invest with.
> I see little reason this means they can turn around a huge beast.
Why do you see little reason? We have repeated examples in recent history of a strong leader turning around what appears to be an immovable object. Jobs at Apple. Nadella at Microsoft. Su at AMD.
I see little reason to believe Intel can’t do the same. They have the technology and history to be successful, they just appear to have had management that wasn’t prepared to deal with a changing market.
I am also glad the Drone CEO is gone. At one point Intel was into drone transport: https://www.youtube.com/watch?v=bFEYDzWJB00 . If they'd focus on semiconductor manufacturing that was slipping under his governance, Intel would be at a better place. And, fixing internal culture.
Same with marketing team. Get back the 90's team that made many memorable adverts. Blueman group, bunny suit ads, the works. Pentium, the word alone brings fond memories. Absolutely legendary. The current Marketing/PR team has created a disaster after disaster (see LTT rants about all kinds of PR issues).
They are basically rebranding their node size to TSMC's ( or more like Samsung's ) naming scheme.
10nm to 7nm.
7nm to 4nm
7nm+ to 3nm
5nm to 2nm
And all of them make senes and has been known for quite some time. I remember writing something on SemiWiki in 2018. So really they are pushing their 7nm to next year, their first node with full usage of EUV. An iteration of 7nm in 2023. I say iteration when in reality that was always the original target of Intel's 7nm. And 5nm in 2024.
10nm is already shipping, and judging by their product lineup seems to be yielding a lot better. Their 7nm next year are all going to be for their Aurora project anyway. So my guess is that it is more like a place holder with little actual capacity for anyone else including Intel's DC and Consumer business. Shipping their own product with 7nm+ / 3nm in 2023. And the real test will be 2nm in 2024. Is it aggressive? Yes, plausible? I am quite optimistic.
>evidence there is that Intel will be able to deliver all this?
Yes. Intel's biggest problem wasn't 10nm, or their chip performance. It was their mentality of keeping Fat profit margin from end product sales. And business School to this day continue to teach the same crap over and over again. That is why having MBA as CEO in tech hardly ever works. The evidence here is quite clear. Intel is really betting on being a Foundry. I do not believe Qualcomm and Amazon will sign up Intel if the price wasn't right. It means Intel is finally doing their second Andrew Grove Moment pivoted Intel away from DRAM to microprocessors, except this time to Custom Fab. As he once said. "Our Strategy is Simple, we build Fabs and we fill them."
Intel should split into two. One a fabless company focus on chip design. The other a foundry business so there is no conflict of interest like Samsung. Apple pull out of Samsung partially because Samsung compete with Apple on smart phone.
if Intel is serious about IDM 2.0. Intel needs to earn their customers trust. TSMC only focus on production and look where they are today.
It seems to have worked out for them? The capital investments required to have a bleeding edge node keeps increasing every generation and there are fewer and fewer companies keeping up over time. AMD just didn't have the volume to keep up. Splitting off the foundry let AMD jump to TSMC for some of their volume and Global Foundries is free to dance with other companies that don't need the bleeding edge.
Intel will still need those x86 profits to catch up with TSMC first. And Amazon is having a head start again. Now I am slightly worried about AMD. May be they will buy Ampere Computing?
Other sources say they have some agreements with ASML. Intel will be the lead customer for ASML's high numerical aperture EUV machines. These will allow Intel to avoid double-patterning. Secondly, both Intel and Samsung are going for GAA while TSMC will still be delivering FinFET.
The capital, agreements and plan to catch up is there. But it's always healthy to be skeptical on execution.
Per the OP, it sounds like they're taking more conservative/incremental steps now, and their previous fault was that they were too ambitious and tried to do too much at once.
My understanding is Intel's rivals also took a more incremental development process, which basically allowed them to capitalize on Intel's missteps.
"But David Kanter, an analyst with Real World Technologies, said Intel is being more cautious than in the past. The years of delays resulted in part from the "hubris" of tackling multiple technical problems in a single generation of technology."
That's some serious spin there. The chatter has been that Intel has had poor focus, poor working conditions, poor investments, poor communication within the company and a plethora of hubris. They need a culture overhaul and that seems to be what's happening. The jury is out to where they take this. I would bet on their competitors at this point.
> The jury is out to where they take this. I would bet on their competitors at this point.
that's a completely uninformed take on the topic. This is like betting on Airbus because Boeing has problems with the 737 - when the reality is there is no circumstance whatsoever where the federal government would allow Boeing to go under.
Intel is already being propped up with some sweet gigs doing supercomputer stuff for the National Labs (that their actual hardware probably does not justify on the merits - much like the national labs throw AMD a bone with CDNA in the GPU compute market) and a bunch of R&D tax cuts. The US also negotiated their way into a TSMC gigafab as a fallback strategy but they very much would prefer the American fab company makes it through as well, and they'll spend the cash to make it happen.
It's far cheaper and faster to get the existing competitor back up to speed, than to build a new foundry company from scratch, or even to just build a TSMC gigafab.
Their competitors also aren't american companies. GloFo is controlled by UAE investors. TSMC is controlled by Taiwan. Samsung is controlled by a Korean chaebol. AMD and Apple are fabless companies. Intel is the only US company still designing chips and producing them at a US foundry on any kind of a competitive level in the performance computing market.
Even Intel cannot be considered as an 100% USA company.
The Israel parts of Intel had essential roles for decades, both in CPU design and in manufacturing, so it is very unlikely that there exists any Intel IP or know-how to which Israel does not have complete access.
The main difference is that Israel is considered as an ally.
7 is half node, 4 full node, 3 half, 20A full node again. Moore's law is doubling of transistors density every 2 years. So the roadmap is indeed conservative, just following the Moore's law.
Indeed. It will be interesting to see how this will interact with architecture roll outs - I wasn't entirely clear how that would work from the webcast.
There isn't any evidence at this point and it could go either way. Though if I had to place a bet, I wouldn't bet against Intel as they will likely have an influx of business/R&D funding from the U.S. govt as well as favorable policies/legislation to help ensure the outcome. In other words, it is Intel's to lose.
Their competition are the ones who played marketing games already, skipping forward a node relative to pre FinFET scaling. Intel's rename aligns it with the rest of the industry in terms of transistor density on each process number.
None of these numbers have aligned with physical features on the chip for a while, so they are just an arbitrary convention now. It's better for all of the manufacturers to use the same convention.
Everybody, apparently except Intel, has already been doing this for years. It's been a long time since process names such as "7nm" actually reflected a minimum feature size of 7nm, and even if it did differing transistor designs means you can no longer predict transistor size/density from minimum feature size.
A much more meaningful way to compare processes across different manufacturers is directly by transistor density, but the "nm" process generation naming tradition is so entrenched that every one is still using it but fudging the number in some sort of attempt to position/align their own processes against those of competitors.
Of course it's not just transistor density that matters, but also power usage and speed, and no one number can capture it all
Makes sense, they have fabs on different process nodes, and if they aren't keeping them running at capacity with their own chips they should be running someone else's design through them. I know, it's not as simple as all of that, but a pivot to serving fables designers as well as their own needs seems like a necessary step as they continue to work out their < 10nm processes.
By opening their factories to external designs they admit they can no longer survive on their own chips manufacturing. It is a new branch of business, but it is a lower margin one, comparing to vertically integrated manufacturing. There's a reason the world has more fabless silicon giants than fabs.
By renaming their technologies they show that the conclusion is that the issue is marketing. Maybe there's an issue with Intel's marketing, but the bigger one is Intel for a while now is not a technology leader. No amount of marketing will compensate that. The slide showing undisputed tech leadership by 2025 is just silly.
Yep, it's a trap, 99%. Intel will be happy to waste lots of money (and investments) to kill wanna-be server CPUs manufacturer. Why earn a bit when you can future proof your possition ?
And morphing into fab ? Numbers renaming is not the same as finally not failing to deliver lower nodes.
The value is really owning some percentage of the foundry market. Only a few can be cutting edge, expensive and time consuming. Better to compete now on volume and cost.
Yet another Intel revolution, where they rotate the business strategy 360 degrees and turn the bullshit screw one more notch on the way to utter irrelevance.
Personally it's sad that we (may) miss Qualcomm's SoCs built with TSMC's latest 5nm and further process. Possible MediaTek's SoC become best perf-watt high-end SoC for Android?
If the agreement is "Qualcomm will send some of its business Intel's way," that isn't a huge vote of confidence. Heck, are we talking about Qualcomm's Snapdragon 8xx series or just some Qualcomm chips like the ones they put in WiFi routers? Or their low-end Snapdragon 4xx series?
The article is kinda spinning it as "Intel will be making the Qualcomm chips going into everyone's phones," but it could be more "Qualcomm is going to have Intel manufacture its low-end stuff which is always behind tech-wise and Intel is giving them a big discount because Intel needs a win."
> In the chip world where smaller is better, Intel previously used names that alluded to the size of features in "nanometers". But over time the names used by chipmakers became arbitrary marking terms... This, he said, gave the mistaken impression that Intel was less competitive.
They are definitely less competitive at this point. Nanometers might not be the right measurement, but one can measure transistor density. Maybe that would make more sense?
Even if "Intel 7" (10nm Enhanced SuperFin) is equivalent to TSMC's 7nm, we saw TSMC's 7nm back in 2018. I have a TSMC 5nm in my pocket. I guess if Intel can keep to its road map, it will catch up. If "Intel 4" comes out with products in early 2023, it won't be far behind TSMC's 3nm process (which will likely debut in the 2022 iPhone).
Of course, we'll have to see how well these Intel chips perform. We should be able to benchmark Intel 7 processors with Alder Lake processors landing later this year. Will these stand up well against AMD's 7nm Zen 2/3? Will they stand up against AMD's upcoming 5nm Zen 4? I guess we'll be able to test Intel's chips against what AMD has been shipping and see how well Intel 7 stands up against both 7nm and 5nm processes.