Intel Honesty

https://stratechery.com/2024/intel-honesty/

by surprisetalk • 2 years ago

376 384 2 years ago

flerchin• 2 years ago

Brutal. Intel processors will be the equivalent of government cheese. Not what anyone wants, but it's what you get with your government dollars.

brcmthrowaway• 2 years ago

I never thought of it that way. Are they going the way of Raythoen/Boeing?

wklauss• 2 years ago

I see the semiconductor industry becoming a bit like the auto industry. A geopolitical pawn, heavily dependent on subsidies and in turmoil due to market forces (switch to electric in cars vs. switch to ARM/RISC in semiconductors)

KerrAvon• 2 years ago

To be clear, the outcome we actually need here is a bulwark against the possibility of China eventually being the world's only producer of leading-edge chips.

datavirtue• 2 years ago

They are building a huge fab in Ohio but I don't see Intel as a going concern that will be around at the date of completion. I fully expect the project to be abandoned.

davidw3• 2 years ago

The geopolitical bit of all this is the real wild card. No one really knows what's going to happen there.

UncleOxidant• 2 years ago

Yeah, you've got Apple, Nvidia and AMD heavily exposed to that geopolitical risk, but seemingly not willing to help lessen it by investing in US fabs. Whenever I note this people here and elsewhere say "they're smart to be fabless" but somebody's gotta run fabs and more fabs need to be run in the US and quickly. I realize industrial policy isn't in vogue these days but it might be a good idea for the future of US semiconductor production to arrange some marriages between some of these companies that have a lot of capital and have a lot of need to reduce their geopolitical risk. Our brand of capitalism isn't great at looking ahead more than a quarter or two.

jjtheblunt5• 2 years ago

Apple is heavily invested in the new TSMC fab in the US east of the north end of Phoenix, last i read; not sure how arranged that marriage really is, but perhaps tax incentives for TSMC facilitated Apple plans to heavily utilize it.

(trying to find linkable articles...like this

https://www.theverge.com/2022/12/6/23497417/apple-tsmc-phoen...

) ; p.s. i wonder if only i compulsively balance lisp parens

sroussey5• 2 years ago

Apple could invest in an Intel Foundry spinoff, but there is no reason to invest in the Intel of today.

Panzer047• 2 years ago

I thought historically companies always wanted "second sources" for things like this.

Surely all the fabless semis can't be blithely running off a cliff with TSMC as their only parts source? That would be a catastrophic risk to take down the track if there's only a single vendor for advanced chips and they've decided to hike prices and collect all of the benefits alone from their advanced processes and share none of them with their customers.

moffkalast• 2 years ago

The geopolitical bit is that TSMC is building local US and EU fabs in case of a Taiwan invasion. Once those are up and Intel becomes redundant the ad hoc subsidies might just dry up.

davidw3• 2 years ago

Sure, but "is building" could be very far from "is able to smoothly switch production if the main plants and some/many of the associated people are war casualties"

hakfoo3• 2 years ago

I have to wonder if a TSMC-owned fab in Phoenix is still a different geopolitical risk factor than an Intel-owned one.

It reduces the deterrence factor of the TSMC facilities on Taiwan: if they get damaged or sabotaged during an invasion, the entire industry doesn't get blown back to 2010. But now I'm imagining they'd have a board of directors having to meet overseas and claim their legitimacy to trading partners, almost like the "governments in exile" during wars.

keepamovin11• 2 years ago

Less so that thought. TSMC is Chinese via it being Taiwanese. No disparagement of regional uniqueness or actual ownership, just its success is a success of the Chinese people (into which Taiwanese are included). China will not substantially harm TSMC when it is so successful at destroying American dominance in a key industry. Taiwan and China are very much aligned when it comes to TSMC.

Maintaining a US native foundry might actually be a way to prevent war because one of the key triggers might be the moment where all major chip players totally depend on TSMC, which gives Chinese/Taiwanese massive leverage. Economically, might be a loss leader. But strategically, important, probably.

andy_xor_andrew• 2 years ago

US chip production really needs its SpaceX moment, but it seems like it will never happen, and we're left with the crumbling empire of Intel.

By "SpaceX moment", I mean a startup entering an impossible market, where the barrier to entry is billions of dollars of research and manufacturing, a market dominated by industry giants from the 60s, and yet still coming out on top somehow.

cherryteastain• 2 years ago

The semi spacex moment already happened - it was when TSMC was founded

hangonhn3• 2 years ago

That's such a good insight. It really was the SpaceX moment or maybe we can say SpaceX was the TSMC moment of space since TSMC came first.

ac29• 2 years ago

1987?

As near as I can tell there was nothing particularly remarkable about TSMC in the 1980s or 1990s.

Wytwwww8• 2 years ago

Or maybe Intel itself? AMD? Nvidia?

milesskorpen• 2 years ago

I think the issue is that TSMC exists, which takes up a lot of oxygen and opportunity. Whereas in space, there wasn't a high quality opportunity, so SpaceX had a lower hurdle to being the best (even though that still wasn't easy!).

doron• 2 years ago

Indeed, but downplaying the strategic threat to Taiwan as a "short-term" potential impact is not a credible position in the long term.

It is prudent, possibly even critical, to have foundries on US shores, and the US will have to pay for it.

cdchn• 2 years ago

Chip production seems another order of magnitude more expensive than space travel, when you look at tens of billions for a fab, but also several orders of magnitude more productive.

nemothekid• 2 years ago

By "SpaceX moment", I mean a startup entering an impossible market,

I think semiconductors is a difficult space for a SpaceX. SpaceX was relatively cheap - the company itself only raised 2 billion dollars, and according to Musk, the cost to develop Falcoln Heavy was "only" $500M. I think the "thesis" behind SpaceX is moreso that the technology to develop rockets had come down massively, but the market had gotten fat and lazy on government contracts. The market only seemed impossible because everyone just assumed so.

On the other hand ASML's EUV machine costs $300M and that's only a small part of what you need build your own Fab and barring some massive research breakthrough that isn't coming down any time soon.

reaperman• 2 years ago

a market dominated by industry giants from the 60s

The thing is, the SpaceX moment occurred because of a rare opportunity where the industry incumbents were insulated from any competition or need for innovation for ~30 years. Some domestic industries still operate in an environment somewhat like this (e.g. the steel industry), but still face modernized global competition (albeit whose effects are kept at bay via import tariffs reaching as high as 266%).

The semiconductor industry... while there are magnificent barriers to entry and not "enough" competition, they have absolutely not been wholly insulated from innovation or competition. And innovation in fabrication is driven at a very rapid pace! So there aren't huge outsized profits to be found like there was for SpaceX. And even SpaceX needed to invent what was almost an entirely new industry (Starlink) to become truly profitable.

And even trying to pull a "SpaceX" on domestic steel manufacturing would fall flat because you wouldn't capture the global market, and would still be dependent on some (potentially much lower) tariffs to compete against the very modernized factories in China, which are innovative/competitive/efficient, unlike our domestic industry.

SpaceX took advantage of a very rare situation where there truly was no competition or innovation anywhere on the planet for a very long period. I'm not aware of any other industries which are quite as far "behind" today as space launch systems were in 2010. But if anyone else is, please mention those industries here!

packetlost2• 2 years ago

US Telecom is not as far behind, but it's pretty far behind. Unfortunately, I think it largely competes with cellular networks and Starlink these day.

highfrequency• 2 years ago

Nice analysis on competitive dynamics of semiconductor industry vs. space industry in the 2000s.

I'm not aware of any other industries which are quite as far "behind" today as space launch systems were in 2010. But if anyone else is, please mention those industries here!

Digital payments may be in a similar situation. Visa's 2% fee may seem low but from a scale and competitive standpoint it is fairly absurd - they are making 80% gross margins on $30b of revenue to do what boils down to a few API calls and some fraud repayment. I doubt that they need to do much innovation to keep that moat either (in contrast to say Apple and their 30% cut of App Store revenues - they need to constantly stay ahead of Android and Windows).

Curious to hear your take on what other industries at least come close to the space dynamics in 2010.

nick238• 2 years ago

I think rockets are insanely simple compared to silicon. The amount of research and development it takes to build a light source (hot tin plasma[1]) is extraordinary, and fabs are also probably the most complicated manufacturing facilities in the world. Tesla struggled mightily with integrating disparate auto parts manufacturers some years ago.

SpaceX and Tesla benefited from being helmed by a crazy person (in their nascent stages), who pushed to break norms in the conventional thinking, either "rockets can't be reused/we must spend $10B on a test campaign[2] before any part leaves the ground" and "EVs aren't cool". I don't think if Musk could design a rocket engine from scratch is relevant, but the strategic design patterns of 1) reduce requirements, 2) remove unused things, 3) simplify/optimize, 4) accelerate cycle times, 5) automate. Those points aren't revolutionary, just a more expanded "go fast and break things."

The computers that came out of Silicon Valley in the late 70s-into the 80s were a disruption to the old stalwarts like IBM. Though for silicon maybe I'm just trapped in that pre-SpaceX thinking.

[1] https://phys.org/news/2020-05-exceptional-euv-hot-tin-plasma...

[2] https://www.planetary.org/space-policy/cost-of-sls-and-orion

kazen443• 2 years ago

The computers that came out of Silicon Valley in the late 70s-into the 80s were a disruption to the old stalwarts like IBM. Though for silicon maybe I'm just trapped in that pre-SpaceX thinking.

another good example of this is how VLSI in particular disrupted the mini-computer/mainframe market and made CPU's cheap enough to put them into smaller and far cheaper machines. This really shook up the old guys who where used to being total system vendors. (IBM, DEC etc). Suddenly, you could get a computer for much cheaper compared to the decade prior, and by the early to mid nineties you had guys like sun eating their lunch completely.

IBM pivoted to being a services company and DEC just imploded.

namibj• 2 years ago

There's another light source that isn't nearly as hard and actually delivers better quality light at far higher electrical efficiency... It's just large and most of the cost is in the large machine that's kept under vacuum. GloFo was working on doing EUV with that (they gave up on EUV), and Japan was or is recently working on a small demonstrator fab with iirc about 10 steppers, starting with the light source machine.

That machine btw is called "free electron laser", and utilizes relativistic electrons to match their velocity to free-space photons instead of the slowed-down-by-metal-helix traveling-wave-tube-amplifiers that power most TV satellites. Those work at about 10 GHz or 30mm wavelength, much larger than the 13nm band used for EUV lithography.

FELs can have efficiencies in the double digit % if built with energy recovery, btw.

FuriouslyAdrift• 2 years ago

TSMC has been spending $15 - $50 billion per year for years to stay on top.

Estimates are it would take $200 - $300 billion just to catch up with them and little to no profitability for a decade while burning $10s of billions every year.

I think the federal govt is the only entity that has that kind of money.

amluto6• 2 years ago

Maybe one of the e-beam startups can pull it off. Aside from scaling issues, direct electron beam lithography ought to outperform optical lithography by many metrics, not to mention that there would never be costs associated with mask revisions.

Here’s one of them: https://multibeamcorp.com/

I remember touring a little research fab, maybe around 2000, that could achieve feature sizes comparable to what TSMC can do today. But they were very, very, very slow.

(Fast-moving electrons are easy to make, easy to aim, and have teeny tiny wavelengths that entirely sidestep most the issues that people have with photons having obnoxiously large wavelengths. But electrons have all manner of downsides that explain why fabs spend many billions of dollars on optical lithography, one of which is that they repel each other, which makes shining a lot of them at a wafer at once quite problematic.)

numpad0• 2 years ago

I think lack of focus on US domestic chip capability is semi-intentional and is an eventuality. Semiconductor industry, and many other manufacturing industries too, seem to flourish at very outer edges of the free world.

It could be simple as the free world not wanting manufacturing capability at home. It could be wanting only grain farms and money printers.

kazen443• 2 years ago

Manufacturing in the west is expensive compared to cheaper developing countries.

Also, manufacturing has a massive impact on the environment, this includes semiconductor manufacturing. It has politically been impopular for a while now.

insane_dreamer2• 2 years ago

By "SpaceX moment", I mean a startup entering an impossible market, where the barrier to entry is billions of dollars of research and manufacturing, a market dominated by industry giants from the 60s, and yet still coming out on top somehow.

This already happened and the startup was TSMC. Unfortunately for the U.S., it's in Taiwan.

keepamovin11• 2 years ago

TSMC would likely never let itself be bought by a Western company. It's too vital to Taiwan's (and China's interest). The best that can be hoped is its ethical engagement with US chip players in decline.

But surely it wouldn't be impossible, with enough capital and initiative, to start a US-side TSMC clone on US soil? Just like the Valley successes of the 70s and 80s prompted savvy Taiwanese investors to found TSMC. We might need to start copying back some of their playbook to chart our own course to success, as they did with ours, then exceeded it with their out strategy and ingenuity, currently.

anitil• 2 years ago

I don't know geopolitics well enough to know the answer to this. Is losing TSMC an existential risk to Taiwan?

Wytwwww8• 2 years ago

By "SpaceX moment", I mean a startup entering an impossible market

SpaceX entered a market with hardly any competition (due to little demand) and is/was extremely reliant on government funding anyway.

Semiconductors are highly competitive and you can't rely on government agencies buying most of what you manufacture.

I'd argue Tesla might be much more closer to "startup entering an (somewhat)impossible market" considering how entrenched non Chinese car makers it was and how hard it was to survive in and especially start from scratch as a small company.

zelias• 2 years ago

I like the proposed approach of purchase guarantees here. Directly injecting cash into Intel just creates a moral hazard where they spend the money dealing with organizational inertia than innovating in the space. Plus, the government could even turn a profit reselling its purchased semiconductors to various US companies!

The US ecosystem badly needs a serious Intel competitor -- because they just ain't it.

jjtheblunt5• 2 years ago

Plus, the government could even turn a profit reselling its purchased semiconductors to various US companies!

No freaking way. How do you think that’s feasible, considering how fast depreciation goes in semiconductors?

wmf9• 2 years ago

Pork bellies also go bad but you can make money trading them. You don't even take delivery.

bryanlarsen3• 2 years ago

The payments are tied to real milestones like building fabs. Intel cannot spend the money on organizational inertia because that wouldn't get them the money.

Invictus0• 2 years ago

It doesn't matter how badly they want to sell the chips if they don't have the money to build the fabs in the first place. That's where Intel is at right now. Building fabs is incredibly expensive and they need the money upfront.

scrlk3• 2 years ago

in the late 2010’s Intel got stuck trying to move to 10nm, thanks in part to their reluctance to embrace the vastly more expensive EUV lithography process

TBH, it's easy to say this with the benefit of hindsight. Throughout most of the 2010s, EUV lithography was like the "year of the Linux desktop" - i.e., this year will be the year where EUV was suitable for high volume manufacturing. I don't really blame Intel for deciding to go with self-aligned quadruple patterning for 10 nm, but combining it with cobalt interconnects was probably biting off more than they could chew.

FWIW, Intel was funding EUV R&D since 1997: https://www.intel.com/pressroom/archive/releases/1997/CN0911...

That press release had an interesting prediction:

Intel projects that the microprocessor of the year 2011 will contain one billion transistors, operating at over 10 gigahertz and delivering 100,000 MIPS (millions of instructions per second).

They weren't that far off in estimating the transistor count and MIPS: the i7-2600k released in January 2011 had 1.16 billion transistors [0] and delivered 117k MIPS [1] @ 3.4 GHz. The clock speed prediction was way off due to the failure of Dennard scaling in the early-mid 00s.

[0] https://www.anandtech.com/show/14043/upgrading-from-an-intel...

[1] https://en.wikipedia.org/wiki/Instructions_per_second#CPU_re...

hangonhn3• 2 years ago

In addition, SMIC in China managed to get to 7 nm without the use of EUV. Also early EUV yields were actually quite low.

wetpaws• 2 years ago

7nm means nothing, it's an arbitrary marketing number

schmidtleonard4• 2 years ago

Did SMIC get 7nm to yield or is it limping but propped up for PR purposes like Intel 7?

api2• 2 years ago

Throughout most of the 2010s, EUV lithography was like the "year of the Linux desktop" - i.e., this year will be the year where EUV was suitable for high volume manufacturing.

We need to learn to recognize the difference between something that's never going to happen for either physical or economic/social/structural reasons, and something that is just really difficult and takes a long time.

I always think of this when I think about fusion and the irritating "fusion is 20 years away and always will be" meme. In reality fusion has been edging closer, and closer, and closer for decades. Like EUV lithography it's just incredibly hard and requires a ton of capital and time and some of the smartest people on Earth to make it a reality. The same logic applies to things like a working HIV vaccine, life extension, or human space flight and space settlement.

The reason there's never been a year of the Linux desktop on the other hand has to do with the full picture of what's required to mass market a desktop OS and support that, not just the technical problems, as well as business reasons around what Microsoft does to incentivize vendors to stay in their ecosystem. Linux desktops are just about ready today but no mainstream laptop/desktop PC vendor is going to sell them for a long laundry-list of non-technical or para-technical (legacy software base) reasons.

SoftTalker• 2 years ago

I'd say Linux desktops have been good enough for a long time, but never broke into the mainstream for the reasons you cite. ChromeOS is the closest thing I guess, but outside of schools and a few other institutional uses, Chromebooks are not very popular.

bee_rider4• 2 years ago

The main reason there will never be a “year of Linux on the desktop” is that desktops became irrelevant before Microsoft became… whatever it is now.

timschmidt4• 2 years ago

Valve sells the SteamDeck and is preparing SteamOS for use on similar handhelds available now from most major vendors.

They may not be traditional laptops or desktops, but they're PCs, and they're being sold with compatibility with a large installed base of Windows games in mind. Moreover, handheld gaming PC users seem to perceive SteamOS as being superior to Windows in this device category.

x0x0• 2 years ago

a working HIV vaccine

It's worth pointing out we kind of have this now. Or, to your point, are definitely inching (a lot) closer. The lenacapavir trial results are great. From the press release https://www.gilead.com/news-and-press/press-room/press-relea...

Gilead’s Twice-Yearly Lenacapavir Demonstrated 100% Efficacy and Superiority to Daily Truvada® for HIV Prevention
– First Phase 3 HIV Prevention Trial Ever to Show Zero Infections

zhobbs• 2 years ago

It’s a good point, but one thing to consider is that ultra long hard tech might be structurally challenging as well. If fusion requires 100 years of capital and R&D it won’t be viable ever due to economic/societal/structural reasons.

bradleyjg2• 2 years ago

…

Never is a very long time, so I’m not going to say never. But it’s not at all certain that when all those smartest people declare victory we’ll have something that doesn’t make economic sense because the upfront capital costs are too high even if the theoretical (at that point) full lifecycle kWh cost is low.

ghaff6• 2 years ago

Intel was also being at least a bit disingenuous in public at the time although they probably also thought they could make those frequencies work. But a very senior Intel exec told me, at the time IBM was showering them with a lot of public snark, that of course they new about the upcoming issues but Microsoft was so worried about multicore scalability that Intel had to play along.

wmf9• 2 years ago

Nobody used EUV for 10 nm though. Even 7 nm doesn't need EUV. This kind of error calls the rest into question.

throwaway484762• 2 years ago

Intel 10nm and TSMC 7nm are equivalent.

encom• 2 years ago

year of the Linux desktop

A tangent, and I realise this phrase is a meme, but it has always bothered me. Nobody has ever defined what year of the Linux desktop actually means, so it's a phrase without any real meaning.

fluoridation3• 2 years ago

Roughly speaking, that running Linux as a desktop OS would be easy enough that even non-technical users would be able to do it with difficulties comparable to, say, Windows XP.

mrsilencedogood• 2 years ago

So did the year of the EUV lithography happen? What year was it?

(Genuinely asking, I don't know anything about chips - I'm the kind of programmer who is profoundly comforted by pretending like my CPU is like a really fast version of the thing I did in CompE 201, as opposed to reality which i gather is a bit more like a demonically possessed stone that glues together a bunch of barely understood physics that kind of "even out" at macro levels but seem pretty f'd up at quantum levels).

scrlk3• 2 years ago

2019, when TSMC N7+ entered high volume manufacturing.

baoluofu• 2 years ago

Your description of reality made me chuckle. I remember during my degree the professor teaching us about how simple CPUs are made, but saying that the people who engineer the current hardware are like dark magicians. It can only have gotten more complicated since then.

PaulHoule• 2 years ago

I remember seeing headlines every 18 months or so that Intel was laying off 10,000 engineers for no good reason. Made me think I didn’t want to go to work for them because I might be next.

AnotherGoodName3• 2 years ago

Intel is trapped. Its debt repayments alone are massive. The poor performance in stock has encouraged 20years of wage stagnation to the point where you can literally earn over 50% more at amd or nvidia or even an startup for equivalent roles so they aren’t hiring the best. Their pay in the Bay Area is double tsmc in taiwan for equivalent roles in raw dollar terms but the ppp differences means your better off working for tsmc in Taiwan than intel in the Bay Area. That’s not a joke. Intel are literally incapable of attracting talent from Taiwan right now.

They don’t have the talent they need and the debt trap and poor performance means a lot of push back to the needed doubling of wages to attract that talent. It’s a very hard sell for any exec trying to correct this problem. They sidelined lip bu tan who was one of the advocates for even more layoffs and wage freezes but he’s one of many backwards thinkers they need to remove. It’s going to be difficult to fix their board.

Without talent intel has no hope of winning and they can’t get that talent due to poor stock performance for the past 20years leading to executives and shareholders wishing to implement the opposite of what they need right now. In fact they have ongoing layoffs right now. A true downward spiral and the only real hope is for a newcomer to step up.

nightski• 2 years ago

I'm not saying you are wrong but personally I'd view the stock as an attractive perk right now since it is so low.

Whereas if I were joining a company such as Nvidia I'd honestly be kind of worried.

Historical performance isn't really a great indicator here.

skeeter2020• 2 years ago

sotck can be low because it's undervalued by the entire market and going to jump once they catch up to the future value potential, or it can be low because it accurately reflects the decline and limited value. If you only see the "attractive perk" from being so low you're the one using historical performance as the indicator, thinking Intel's only a few quarters or years away from former glory days.

AnotherGoodName3• 2 years ago

You can trade your stock for intel stock when working at any company. Stock grants really should be swapped for income and reinvested how you see best. So ultimately the total take home income is what should matter for any new hire. Take the higher paying and job and buy intel stock if you believe it’ll rise.

packetlost2• 2 years ago

I mean, I have a lot more confidence that nvidia will continue to deliver products to a market that wants to pay for them over Intel at this point in time, even if the stock is likely to come crashing back down to earth soon. As an employee, I wouldn't really factor in stock performance necessarily, but the overall image and outwardly visible struggles Intel is going through tells me the internal struggles are probably far worse than the public or shareholders know.

bryanlarsen3• 2 years ago

By what metric do you consider INTC to be low? I took a look because I'd be happy picking up some cheap INTC, but it still looks expensive to me.

ohcmon• 2 years ago

You would be surprised, but nvidia’s employee stock plans allow to select the purchase price within the last 2 years https://www.nvidia.com/en-us/benefits/money/espp/

bluecalm3• 2 years ago

Intel stock is as high as it was 15 years ago when they had total market domination. Now they are at the brink of collapse. The market is way bigger today but still, I would take a dominant player over a failing one every day.

Another way to look at it: Intel market cap is 83B, AMD's market cap is 228B. Do you think Intel is expected to make 1/3 of the money AMD is going to make in coming decades? I see no reason to be as optimistic.

dv_dt• 2 years ago

I think this contains mostly correct analysis with respect to hiring experienced talent, but under emphasizes how unique each process is per company. Experience is a positive, but less positive because of it.

There is a lot of potential that could be made from newer hires combined with focused in house training/experimentation - which is what all these businesses had to do in their initial expansion - with not enough people of experience at scale available, and they probably had to do a running training ramp up multiple times over multiple generations of growth. This is in general is an underutilized strategy - esp for mature companies that need to essentially build a new generation of tech in-house.

FredPret• 2 years ago

It's grim at INTC but don't forget the CHIPS act and whatever else will follow that. The US government spends 0.5T per month - more than the market cap of all but the top 15 American companies [0] [1]

Not saying it's a good or bad idea, or that it will or won't happen - but if the US government decides to reinvent Intel, they can easily write a cheque that (if spent wisely) might do the trick.

[0] https://companiesmarketcap.com/usa/largest-companies-in-the-...

[1] https://en.wikipedia.org/wiki/United_States_federal_budget

twobitshifter• 2 years ago

Intel is underserving of this investment and the chips act has no teeth to ensure that they follow through and build these facilities and create jobs. They will not follow through and spend the money wisely and that was clear from the moment that bill was drafted.

kridsdale3• 2 years ago

Why doesn't the USA just buy out all of Intel's debt, let them start fresh? Call it a Silicon Era War Bond, in reverse.

trhway• 2 years ago

It's grim at INTC but don't forget the CHIPS act and whatever else will follow that. ... if the US government decides to reinvent Intel, they can easily write a cheque that (if spent wisely) might do the trick.

I suspect the result of the CHIPS money going to Intel would be Starliner from Intel (btw, Starliner did meet all the development milestones set by the government, yet here we're). Rotten overbloated corporate managerial bureaucracy can digest any amount of money you throw at it, and it will make it only more rotten and overbloated.

joe_the_user4• 2 years ago

I worked for Intel in the 90s.

I think the company always paid "industry standard" compared to other companies paying actually well. And it was a terrible place to work by most objective measures - people got let go quite freely, etc. When Andy Grove writes a book called "Only The Paranoid Survive", he's not committing to employee loyalty.

I think the way Intel got good people, and they did get good people, was by a combination of the opportunity to build something that gets widely used and a cult-like spirit of "are you good and tough enough to survive the bullshit".

But yeah, seems there's no recovery when that approach stops working. And it's disturbing that many of America's "crown jewels" (Intel, Boeing, etc) are basically constructed that way.

jeffparsons• 2 years ago

I think if Intel was a private company then it would have a better chance of recovery via a collection of focused experiments to overcome their biggest technical deficiencies (compared to their competitors) or to, e.g., figure out a compelling new product that the market didn't realise it needed, but that doesn't require solving difficult physics problems.

But to do this sort of thing you need a dictator at the top who is willing to risk a run of negative-profit quarters to fix the company's underlying rot. If you try to do anything like that as a leader of a public company, then shareholders tend to get angry.

I wonder if there's any lesson that could be distilled from the (minority of?) public companies that don't end up settling into a pattern of carefully-managed mediocrity. Is there a unifying theme? I haven't spent enough time thinking about this to even propose an answer other than "cult of personality around the leader" as maybe helping.

theevilsharpie• 2 years ago

But to do this sort of thing you need a dictator at the top who is willing to risk a run of negative-profit quarters to fix the company's underlying rot.

Private companies still have shareholders that the CEO answers to -- who tend to get angry at taking losses quarter after quarter with no clear path to growth.

bradleyjg2• 2 years ago

They sidelined lip bu tan who was one of the advocates for even more layoffs and wage freezes but he’s one of many backwards thinkers they need to remove.

Genuine question: why isn’t what they need large layoffs and dramatic wage increases?

y-c-o-m-b• 2 years ago

I worked there for a few years a little over a decade ago in Oregon. The wages were abysmal, so you're not wrong, but they still drew a lot of good talent because Intel was more accessible to the suburbs than the city of Portland (commuting to downtown Portland was awful) and remote options were limited. Krzanich era layoffs and the availability of more remote work options resulted in a rapid drain of talent though.

The working environment was toxic (still is based on what I've heard from folks I know working there). It encouraged working against your peers instead of with them. Top heavy management meant lots of poor choices and money burned on worthless projects that never saw the light of day. Always had the looming threat of layoffs which ate into morale. TONS of off-shoring and third-party contracts, which everyone here knows is a mess in itself. Piss-poor research before buying up smaller companies or doing joint-ventures. Pouring money into the construction of more buildings when so many of their current ones sit unoccupied. It's all just a recipe for immense failure.

svnt• 2 years ago

I worked there on a contract while they were building a phone and had as many as six managers over six months. One of them would come in at 11, talk to me about nothing until lunch, then come back after lunch, talk more, and go home. I had to come in early to get anything done.

When I was toward the end of my project and sure I wasn’t going to try to get a blue badge I laid into someone in a meeting. It wasn’t my best moment but I had absolutely had it with this person from a competing group blocking my progress. I was on my way out anyway. I wasn’t sure if I would be fired.

The manager from the competing group sought me out the next morning to offer me a L6/L7 position. He said he wasn’t sure I had what it took until that meeting.

My group manager did the same later that day.

I made the right decision leaving.

heresie-dabord• 2 years ago

Your second paragraph, with a slight reworking, is a succinct diagnostic for all bloated and dying corporate entities:

"Diagnostic for immense failure. Does the organisation...

= enable a toxic working environment that is openly discussed,

= encourage internal conflict act every level, business unit and staff,

= cultivate top-heavy management, promote low-talent that makes poor choices and burns money on worthless projects that never deliver,

= maintain the threat of layoffs to destroy morale and engagement,

= send critical functions off-shore and/or to third-party contracts,

= fail to do strategic research, make poor choices in acquisitions and partnerships, and build unnecessary facilities?

If so, the patient manifests late-stage corporate morbidity."

keepamovin11• 2 years ago

If AAPL bought Intel could it fix it?

keepamovin11• 2 years ago

Do you work in the Taiwan side of the industry?

HDThoreaun• 2 years ago

Intel's debt is not bad at all. My reading has it at $53 billion. Before their revenue collapsed in the last few years that debt is only about 50% of revenue. High but manageable compared to att or warner discovery who actually have unmanageable debt. Intel's problem is that revenue is decreasing, and quickly.

TheAmazingRace• 2 years ago

I'm still hopeful of the fact that Intel is, in many aspects "too big to fail", and with their cash on hand, they have enough of a burn rate to turn this ship around, even if they aren't being as intense on cutbacks as they should be.

The question remains, will Intel survive in its current form or could an activist investor stir up a hostile takeover and change the calculus?

mnau3• 2 years ago

On the other hand, Intel is small enough to rescue. $82bn (market cap) is not the amount of money it used to be.

linuxftw• 2 years ago

Intel has $29B cash on hand. Their new fabs are being subsidized. Sure, American labor is more expensive than APAC, but I don't see that being a huge differentiator in such an automated manufacturing segment.

My personal rabbit-hole conspiracy is that AI is driving a fire under the national security apparatus, and we're going to see a restriction on AI-chip technology being exported or manufactured overseas. Intel will be in prime position to onshore that manufacturing.

crowcroft3• 2 years ago

$29b cash and about $50b in debt. They have a valid business still, but I think you're under appreciating how much of a hole they need to dig themselves out of.

Their current foundries can't just make other chips, and they certainly can't make GPUs for AI (TSMC make Intel GPUs). They are making a new foundry in Ohio, but in Intel's current state and comments they've made to the market it's not guaranteed the foundry will be built.

Intel might make it out of the hole, but it's going to get worse before it gets better. At their current trajectory their is not valid reason for the company to have 100,000 employees. I expect the recent layoffs are the first of a few.

rstuart4133• 2 years ago

$29b cash and about $50b in debt.

I know nothing about finance, but I have trouble figuring out how they got themselves into this situation given this https://www.intc.com/stock-info/dividends-and-buybacks :

    As of June 29th, 2024, we were authorized to repurchase up to $110.0 billion, of which $7.24 billion remain available. We have repurchased 5.77 billion shares at a cost of $152.05 billion since the program began in 1990.

Had they not bought there own shares they would be sitting on $70B cash and no debt.

bee_rider4• 2 years ago

I wonder if Global Foundries will become relevant again on the high end. They “gave up,” but they are working on some pretty small nodes now, 12nm, obviously not what TSMC is up to, but maybe they’ll catch Intel in 10 years. I’m sure they are fine for automotive.

onepointsixC• 2 years ago

They’re 4-5 nodes behind Intel today, with no EUV machines let alone High NA EUV. They’re just not relevant in leading edge in the slightest.

closeparen5• 2 years ago

Their pay in the Bay Area is double tsmc in taiwan for equivalent roles in raw dollar terms but the ppp differences means your better off working for tsmc in Taiwan than intel in the Bay Area. That’s not a joke. Intel are literally incapable of attracting talent from Taiwan right now.

There's some whiplash in hearing constantly how the US is a corporate-dominated oligarchy, and then also seeing trillion-dollar industries at the forefront of the global economy constantly getting their shit rocked by a few dozen NIMBY retirees at city council meetings.

amluto6• 2 years ago

Some of the issues are way downstream of those NIMBYs:

Non-housing costs are quite low in Taiwan. Food and childcare, in particular, are so much cheaper than California that it’s hard to believe.

But the NIMBYs aren’t totally in the clear. Rental housing with 3+ bedrooms (for families) is severely lacking in much of the US. Maybe fire codes are to blame. Taiwan is full of very nice new high-rise development that contains units with lots of bedrooms. Wandering around those developments and the new 3-story developments in California, the ones in Taiwan are much, much, much nicer, even from the outside. The NIMBYs should take note.

closeparen5• 2 years ago

While there are a number of technical issues with respect to the limited trickle of multifamily permits that are given, the fundamental dynamic is that engaged voters don't want to see significantly more people living in the region, and therefore it doesn't happen.

lotsofpulp6• 2 years ago

Part of the equation must be the simple fact that the US is big and has the option to give many people the 0.1+ acre detached single family home with a yard and 2+ car driveway/garage lifestyle (and schools with more exclusive student populations).

In smaller places, that simply isn’t an option, so there exists greater demand for family living in high rises.

colechristensen5• 2 years ago

Rental housing with 3+ bedrooms (for families) is severely lacking in much of the US. Maybe fire codes are to blame. Taiwan is full of very nice new high-rise development that contains units with lots of bedrooms. Wandering around those developments and the new 3-story developments in California, the ones in Taiwan are much, much, much nicer, even from the outside. The NIMBYs should take note.

And it's because of zoning. Cities allow density construction but then it's almost exclusively 0 and 1 bedroom apartments because you can charge more per sq ft. Zoning laws should force apartment construction to include multiple bedroom units but that would lower the cost of 1br units. So NIMBYs don't want density construction, and the people who build density don't want big units.

So cities are unlivable if you're not young and single or happy living in relative squalor. So people either don't have children or have to move to have children and you get cities that push out a big chunk of the demographic and then force people to commute.

geodel2• 2 years ago

All true. The question still remains how many US folks can /want to go live in Taiwan vs How many Taiwanese can / want to live in US suburban dystopia. And I say this as immigrant to US increasingly disenchanted by suburbia.

From what I read housing price to income ratio has increased ~2.5 times from 6.5 to 15.8 in last two decades in Taiwan. And it seems to be worse in new developments in city like Taipei.

At a top echelon of Taiwan society or as a tourist it must be nice live, roam around in pleasant urban environments compare to US suburbia but I am not sure average Taiwanese are finding life great with stagnating wages and unaffordable housing.

kilotaras• 2 years ago

Non-housing costs are quite low in Taiwan. Food and childcare, in particular, are so much cheaper than California that it’s hard to believe.

Those are downstream of housing restrictions to a large extent. From "Housing theory of everything" [0]:

Consider a cleaner living in Alabama. In 1960 they could move to NYC and earn wages 84% higher, and still end up with 70% higher income after rent. In 2010, they could move to New York City and become 28% more productive, and earn a wage 28% higher – and reduce the surplus of workers back home, letting them demand higher pay. But since housing costs are so much higher, the net earnings and living standards of someone like this would fall if they moved today, and wouldn’t be worth it. The same would be true for plumbers, receptionists and other professions that allow other people to specialise at what they’re best at and minimise the time they spend on things like DIY and answering the phone. By contrast, top lawyers get wage boosts that are still sufficiently higher to justify a move in both 1960 and 2010, even after the higher rents they’ll have to pay.

[0] https://worksinprogress.co/issue/the-housing-theory-of-every...

qwytw3• 2 years ago

Taiwan is full of very nice new high-rise development that contains units with lots of bedrooms

But presumably you'll never be able to buy and will have to rent forever? Real estate is significantly more expensive in Taiwan relative to income than in SF/etc.

pchristensen• 2 years ago

The primary problem for getting 3+ bedroom apartments built in America is school funding. School districts are mostly locally funded (sometimes with a portion funded by the state government) and almost entirely from property taxes. At a very high level, schools are funded by the acre and their expenses scale with the 2nd-5th bedrooms in their district. A 4 bedroom apartment or condo and a 4 bedroom house will usually pay very different tax rates, so local governments are highly incentivized to deny and obstruct large apartments. A 2 bedroom apartment with one child is one thing, but a 4 bedroom apartment with 3-5 children is a huge money loser.

That's also an obstacle to a lot of cities densifying and intensifying. Built out cities have a model that lets them operate, but not capital or land to build and expand in areas of high growth and demand. That's just one of many factors that are freezing our cities in amber.

zelon88• 2 years ago

Non-housing costs are quite low in Taiwan. Food and childcare, in particular, are so much cheaper than California that it’s hard to believe.

That statement is the problem.

Nobody wants to rent. There are too many rental properties, and not enough affordable housing. The reason you need rental properties is because you're already trying to make a living from families having a home. You're not supposed to do that. That's supposed to be the prize for the family for assimilating with capitalism. By buying it and renting it back to them you're disincentivizing the working class from working.

But by all means, keep privatizing the shit out of 3 bedroom family homes and see what happens. Take every drop of value out of the housing market and bank it. That's what pitchforks are for.

ak2174• 2 years ago

Absolutely true. Also I find it strange that the companies in question could easily resolve their workforce issues by forming a fund to build better/faster public transit to link up their offices with less NIMBY-dominated cities, but choose not to do so.

kbolino4• 2 years ago

The federalized nature of American government means both things can be true simultaneously, even though they seem paradoxical when placed together in juxtaposition. The Feds have supremacy in certain matters, but not all.

Moreover, lobbying is more a question of connections and relationships which are lubricated with money than pure spending power, so it can be easier for a large corporation to nudge things its way at the national and state levels while still struggling to curry influence at the local level, and vice versa for small companies.

insane_dreamer2• 2 years ago

trillion-dollar industries at the forefront of the global economy constantly getting their shit rocked by a few dozen NIMBY retirees at city council meetings

A lot of Intel R&D (and all their manufacturing) is outside the Bay Area, i.e., Oregon and elsewhere, so not sure this is a factor in their case.

kmeisthax2• 2 years ago

The thing that makes America a corporate-dominated oligarchy is veto rights. The American middle class has enough political power to get issues on the table but not to overcome an anonymous rich person saying "no". NIMBYism is the same underlying power - vetocracy. So corporations would have to spend way more time fighting their own power to get things done.

closeparen5• 2 years ago

This is an insightful synthesis, thank you.

dang• 2 years ago

We detached this subthread from https://news.ycombinator.com/item?id=41448341.

ars• 2 years ago

Did I misunderstand or did the author advocate a world where the only high-end chip manufacturer in the entire world is TMSC?

Just the one single company, located in a politically unstable area?

And putting aside the politics, if they become a monopoly what do you think will happen next?

randerson• 2 years ago

Purely from the perspective of Intel and their shareholders it makes sense. But it's obviously a terrible idea for the world at large. TSMC and ASML need US based alternatives.

manav• 2 years ago

Intel was getting something like $20 billion from the CHIPS act - but it seems like by the time the fabs would have been ready TSMC will already have them beat to 1.4nm/a14 (also using High NA EUV). Intel has just consistently failed to execute whereas TSMC has been fairly close to schedule.

newsclues2• 2 years ago

What happens to the computer and tech industry if Intel fails?

The x86 licence still has value to some others in the space, but it seems like Intel could potentially be sold off for parts if it can't sell correct.

_w1tm4• 2 years ago

ARM wins the instruction set wars and buys the tech. Future ARM processors will have x86 hardware that offers compatibility with legacy software. Eventually the instruction set is forgotten.

kbolino4• 2 years ago

Why/how would the death of Intel kill off AMD too?

magicalhippo• 2 years ago

Got me thinking, from what I've read, modern high-performance cores do a lot of similar tricks and translate everything to uops anyway.

How hard would it be for say AMD to make a performant dual-ISA CPU?

I was thinking somewhat like how virtual 8086 mode[1] worked on 32bit CPUs, ie OS could be ARM-based, but it could run x86 processes.

I assume it would be hard to not sacrifice performance for one of the ISAs, but if it's for legacy applications you wouldn't need top speed necessarily.

Are ARM and x86 just too different to make it work? Are there other obstacles?

[1] https://en.wikipedia.org/wiki/Virtual_8086_mode

blasphemers2• 2 years ago

Not sure how much x86 is actually worth considering x86-64 is owned by AMD

fngjdflmdflg• 2 years ago

As another commenter pointed out, AMD owns the x86_64 patents. The original x86 patents have long since expired already, and the x64 patents should be expiring soon as well. The ones listed on Wikipedia have already expired[0] but it is not clear how many extensions are needed to run most modern x64 binaries and if they can be easily emulated via another extension or through microcode. (if that is even allowed).

[0] https://en.wikipedia.org/wiki/X86-64#Licensing

(From the above link, US6889312B1 actually had its patents mistakenly transferred to Philips which seemingly went unnoticed for 18 years)

015a2• 2 years ago

I'm not sure if I buy the thesis entirely (that thesis most aptly being: "Gelsinger does have one fatal flaw: he still believes in Intel, and I no longer do.")

The biggest reason is geopolitics: If escalated confrontation happens with China, Intel becomes one of the most valuable companies on the planet, period. TSMC will finish western fabs eventually, but the delta-t on when they become competitive (tech, capacity & cost, remember) with even Intel's western fabs is... decades, plural? Maybe never? The CHIPS act helped, but TSMC is dragging their feet for very, very (existentially) good reason. Also, remember that TSMC is built entirely on the back of ASML; conflict with China doesn't just block the west's access to the east's fab capacity, it blocks the east's access to the west's tooling to build more and better fabs. TSMC craters in this scenario. Samsung also exists. That's the list of near-SotA fab capacity, globally.

Capacity is another good reason: TSMC is not a bottomless bucket, and every year for the past at least two years Apple has purchased 100% of their SotA fab capacity. No one is competing with Apple's margins (except Nvidia, but that's a bubble market; numbered days). His argument is that there's no market reason for Intel's fabs to exist; we don't have the data to say for certain, but I'd guess that if Intel went to TSMC and said "you're already making lunar lake, make our xeon chips too" TSMC would say "we can't" (especially on SotA nodes, but maybe even near-SotA). They're tapped out, they grow, they're instantly tapped out again, everyone wants what they're selling. Intel fabs Lunar Lake and Arc with TSMC; both very-low volume.

Also worth keeping in mind: Intel was at one time the global leader in chip fabrication; but they lost that crown. People view TSMC as this unassailable beast, but they're just as fallible; and when I hear people say "Well, Intel 18A is probably three years out and by that time we'll have TSMC N2P with backside power delivery so who will even care about Intel 18A" are just extrapolating history. That's a dangerous game when all these rankings and valuations are based on asymptotically approaching the limits of the laws of physics. And while it is unlikely that Intel will take the lead again, TSMC showing any sign of faltering will raise the relative value of the #2 companies.

Intel is not in a good spot, but they're still an interesting business, they're still designing some of the best chips on the planet, and with the right decisions could grow to be even better than interesting. IMO, this article is missing a lot of the hard analysis and data that Stratechery is usually known for; I don't feel you can have a researched discussion on Intel without talking about fabrication volume or their concerning levels of debt, but he didn't mention either of those things. Heck, more than one passing mention of China feels kinda important to the topic.

B1FF_PSUVM• 2 years ago

remember that TSMC is built entirely on the back of ASML;

Probably a few people east of the Netherlands have noticed and are doing something about it.

astrange12• 2 years ago

That's not the only thing you have to replace. ASML has layers of sole suppliers themselves, like Zeiss who make their lenses.

EUV itself was developed by US government labs; ASML is a sole supplier of it because we licensed it to them and refused to give it to Nikon/Canon for no good reason.

pphysch2• 2 years ago

The biggest reason is geopolitics: If escalated confrontation happens with China

If escalated confrontation happens with China, and that is massive IF that is dependent on insanely belligerent action from the West like giving nukes to Taiwan, then new semiconductor manufacturing will be the least of our worries. A Chinese blockade on Chinese and East Asian exports to US would result in total economic collapse in USA over a relatively short period. $100,000,000,000s in real demand for goods no longer being met over the span of weeks.

The possibility of direct conflict with China is the domain of armchair strategists that consume too many Reddit headlines, Peter Zeihan, Economist, etc., and individuals directly downstream of the defense budget, where Threat Inflation is good business. It's really not something that serious, intelligent people are anticipating on the horizon (again, outside of the wildcard of an insane new administration in Washington). Even the insane combo of Matt Pottinger, Mike Pompeo, John Bolton didn't get close to manifesting it (allegedly, some of their attempts were undermined by backchannel diplomacy from Gen. Milley and the JSOC). So what's the real risk?

015a2• 2 years ago

There are a thousand different ways that "escalated confrontation" with China can fall short of "direct conflict" while still negatively impacting TSMC's ability to allocate production to the west. That's why I said "escalated confrontation". If the only ways you can think of it happening is what Zeihan says, then that's simply a failure of your own creativity.

One moderately realistic one: Taiwan experiences something similar to what happened in Hong Kong, during a time when US leadership is not motivated to aid in its defense. This isn't a situation where China and the US are at war; we're still trading with China; and broadly speaking the economies of both countries are fine. But, Taiwan isn't, and whether it surfaces as key TSMC talent/tech leaving the country, military action, or just PRC subtly shifting TSMC's allocation priorities toward the homeland: western chip manufacturers could benefit. This isn't seemingly likely to happen in the next four years given the stated priorities of both sides of the political isle; but the right is shifting more-and-more isolationist every year, and no one arguing in good faith would speak with such certainty as you have on what happens 2028 and onward.

mrighele• 2 years ago

that is dependent on insanely belligerent action from the West like giving nukes to Taiwan

The USA administration has stated several times that they would defend Taiwan if China invaded (see for example [1]). Would you consider that an "insanely belligerent action from the West"?

You can argue that China is not going to invade any time soon, but I wouldn't consider the probability negligible.

[1] https://thediplomat.com/2024/06/what-to-make-of-bidens-lates...

TheAlchemist• 2 years ago

While I agree that a full blown confrontation with China is unlikely (although I do think the probability of it is nowhere near 0), I think your thinking is kind of flawed too. Why would there be a total economic collapse ?

What are the strategically important things imported from China that US couldn't live without, and that couldn't switch to importing from eslewhere or manufacturing at home, in times of war ?

My understanding is that US would is importing a lot of non critical things, whereas China is much more dependent on importing critical commodities (food, oil).

rl3• 2 years ago

(allegedly, some of their attempts were undermined by backchannel diplomacy from Gen. Milley and the JSOC)

Did you mean JCS (Joint Chiefs of Staff)?

cs702• 2 years ago

The OP makes a compelling case that Intel's foundry operations have fallen so far behind that they are no longer economically viable on their own, in the face of current market forces, i.e., Intel's foundry business may not be able to earn a positive return on the tens of billions of dollars of capex now required to catch up with TSMC.

If the US truly views having a domestic foundry as critical to national security, the US federal government has no choice but to pay up big time to support Intel's foundry business until -- hopefully, eventually -- that business is able to compete profitably against TSMC to manufacture chips for Apple, Nvidia, AMD, etc.

Oh, how the mighty have fallen!

cogman103• 2 years ago

There are a slew of foundries in the US already that aren't intel and are open to manufacturing for non-intel products. For example, Global Foundries.

I just don't see what benefit at this point the US government would see investing in intel when they aren't really a whole lot better at the game at this point vs pretty much anyone else.

wmf9• 2 years ago

Intel is a decade ahead of other US foundries.

kazen443• 2 years ago

Also, unlike TSMC, which the taiwainese goverment sees as a national asset and which actually garuantees them economic co-dependence with the west and thus independance from china,t he US seems to have forgotten that having your own manufacturing capability is very important for national security.

Panzer047• 2 years ago

They aren't competitive right this moment, I suppose. Is that a reason to completely give up on all of the embedded knowledge inside the Intel IFS organisation?

Surely at some point you can make the argument that it's worth the investment for possible returns in the future, even if it costs a crazy amount of money right now. Trying to replicate what Intel has as another player would be even more exorbitant, presumably.

jhallenworld3• 2 years ago

The Altera merger pain is maybe enlightening.

Altera 10-series FPGAs were massively delayed due to Intel's 10 nm fab problems. But I speculate that there was also a big difference in toolchains, does anyone know for sure? I mean Altera was using TSMC previously, and I presume were using industry standard tools: Cadance / Synopsis. But I would guess Intel was running their fab on home-grown tools... what is the status these days? For example I know for sure that IBM's synthesis was "Booledozer".

It's interesting because maybe Intel will spin out the fab business. But are they immediately ready to become a commercial fab business instead an Intel-only business? What's the toolchain like for Intel fab?

wmf9• 2 years ago

Historically Intel did use non-standard EDA tools which is one reason they had trouble getting outside customers onto 10 nm (besides the fact that 10 nm didn't work). Some Intel acquisitions like Fulcrum and Barefoot never used Intel fabs.

I think they're supporting a more standard toolchain starting with 18A.

jhallenworld3• 2 years ago

I see, this is their newest node or close to it.

https://www.anandtech.com/show/21504/intel-18a-status-update...

"For Intel, getting an external PDK out for a leading-edge process node is no small feat, as the company has spent decades operating its fabs for the benefit of its internal product design teams. A useful PDK for external customers – and really, a useful fab environment altogether – not only needs process nodes that stick to their specifications rather than making bespoke adjustments, but it means that Intel needs to document and define all of this in a useful, industry standard fashion. One of the major failings of Intel’s previous efforts to get into the contract foundry business, besides being half-hearted efforts overall, is that they didn’t author PDKs that external companies could easily use. "

But some bad news today:

https://www.reuters.com/technology/intel-manufacturing-busin...

arder• 2 years ago

The Altera merger was a perfect storm. I could write for pages about what went wrong. But here's a few key things: First FPGA companies have been massively hit and miss with their product - they were perfectly capable of screwing up their next chip all on their own (it's a mutually beneficial duolopy of crapness with Xilinx). Second, when Altera agreed to fab with Intel the deal was written such that Intel could never buy any FPGA company other than Altera, so from day 1 this wasn't "Let's produce this together" it was "Let's line up Intel to buy Altera" and so Altera never had to deliver anything, they just had to wait for the acquisition. Third, Altera were one of Intel's first customers and it turns out that Intel's fab process couldn't make some of the gates that Altera needed, which is why in the end they partitioned off the entirety of the transcievers to a separate tile fab'd by TSMC and sellotaped to the edge of the Stratix 10.

Also just to check your facts on the 10nm fab problems. Intel's first FPGA for Altera was always planned to be 14nm. It was totally trash for reasons entirely within Altera's control (don't rock the boat, just wait for the acquisition to close). And the synergistic products were a dead end. I'm sure the 10nm catastrophe didn't help but really that time would've been well spent unpicking the disaster of Stratix 10.

jhallenworld3• 2 years ago

Oh right, 14 nm.. it was so long ago now :-)

Also I assume that both AMD and Intel bought FPGA companies for FPGA co-processor acceleration opportunities. I don't think this has panned out.. I know Microsoft has their bing FPGA accelerator, but I don't think FPGAs have made any dent in the AI space, and are certainly not essential in any of the hyperscalers. Maybe AMD will sell Xilinx at some point.

they partitioned off the entirety of the transceivers to a separate tile fab'd by TSMC and sellotaped to the edge of the Stratix 10

awesome.. I could imagine it was too risky to redesign the transceivers.

hash872• 2 years ago

Not a trade lawyer but as I understand it, WTO treaty obligations specifically forbid the kind of purchase guarantees that Ben's arguing for here. Part of the whole point of the WTO is to to pull up the ladder and prevent countries from doing what Japan, South Korea, and Taiwan successfully did to become rich. So this kind of support for domestic manufacturers is not allowed

synergy20• 2 years ago

In June 2021, Gelsinger named 22-year Intel vet Sandra Rivera to head up a new Datacenter and AI Group. Rivera, who spent two years as Intel’s chief people officer before taking on her new role, describes what Intel staffers are looking to get out of their relationship with the company, using terms that would have been unlikely to come up in its original heyday. “For sure, we are in an age now where our employees are looking for a deeper, richer connection to the purpose of the organization,” she says. “They care about diversity, equity, inclusion. They care about the planet.”

throwaway484762• 2 years ago

That does not sound like the constructive confrontation of intels heyday.

umvi2• 2 years ago

today AMD has both better designs and, thanks to the fact they fab their chips at TSMC, better processes

On paper AMD is better, but in the scientific community, it seems that Intel has much better performance for things like NumPy and SciPy. The reason seems to be "Intel® MKL" (https://en.wikipedia.org/wiki/Math_Kernel_Library).

I'm using a lot of weasel words like "seems to" because I haven't rigorously proved it. But anecdotally, my company's AI pipelines which are NumPy/SciPy heavy run an order of magnitude faster (2 seconds vs. 20 seconds) on my laptop's Intel i7 than the do on my Ryzen 7, despite the Ryzen 7 being a newer gen than the i7.

wmf9• 2 years ago

MKL detects non-Intel processors and uses a slow code path.

sliken• 2 years ago

Indeed, and ignores the CPU feature flags for things like AVX2.

There is various hacks, some to the binary (to edit out the if <AMD> then slow) or even just ENV flags. This fixes things like matlab's use of MKL and shows a significant performance increase.

eigenspace• 2 years ago

But anecdotally, my company's AI pipelines which are NumPy/SciPy heavy run an order of magnitude faster (2 seconds vs. 20 seconds) on my laptop's Intel i7 than the do on my Ryzen 7, despite the Ryzen 7 being a newer gen than the i7.

I very very seriously doubt this is a like-for-like comparison. I suspect one case is multithreaded and the other not, or that one is using generic non-SIMD fallbacks or something like that.

umvi2• 2 years ago

All I know is, when I run `python3 ai_script.py`, it's much slower if I invoke that command on my Ryzen 7 desktop PC vs. my i7 laptop. I briefly profiled it once, and it seemed like `skimage.measure.label` ran much slower on AMD.

benreesman4• 2 years ago

I’ve heard the theory that Intel is physically incapable of doing the lithography transition they need to do without missing at least one full release cycle. Which should be bad news but shouldn’t be existential.

The theory goes that any CEO who does the necessary thing will be fired, so either they won’t do it, or the board will replace them with someone who won’t do it.

That kind of market failure can destroy arbitrary value.

wmf9• 2 years ago

They skipped a release cycle with 14th gen (it's the same as 13th gen) and the lithography transition is basically finished already. It looks like Intel 18A will come out before TSMC N2.

benreesman4• 2 years ago

I hope you’re right, I’m not an Intel fanboy but when they hit a high note it’s usually pretty good, and we like competition.

I had a 12900K that was one of my favorite chips in ages but nothing since then has looked appealing.

cbanek• 2 years ago

Sidequest:

“I’m going to do Windows and then I’m going to do Windows Mobile and I’m going to do Windows embedded.” (all on intel x86)

As someone who worked on Windows Mobile and Windows Embedded, it's funny because that's what management was hoping, but it was so different. I don't think any of the phones had an x86 chip in them, although that was our debugging setup so we could use vastly more powerful desktop machines as devkits (although we still had devices to). Having to port parts of Windows to ARM and even MIPS was a giant PITA.

Windows Embedded is closer since it was based on XP, but also a completely different branch from what I remember from Win Mobile.

A lot of people think it's easy to port things, and hard to write things. With really, really complicated software, it's still really hard just to port things.

benreesman4• 2 years ago

I generally like Stratechery but this is not up to the usual standard.

To the extent that CISC means anything it means “backwards compatible”. Modern ARM cores have multi-step decode and issue: to the extent that RISC means anything it means “we developed this with twenty year’s hindsight so we don’t have to do a lot in microcode yet.” In die area, in TDP, in wafer consumption, it’s just not a big cost.

To the extent that Intel missed AI so did everyone: right now there’s one target and that’s NVIDIA, no amount of HBM/TC ratio tuning is going to change that. Like 3 of 17 Ray/Anyscale tutorials target Habana. The Gaudi stuff works fine if you’re willing to pay the “last year’s model” tax all non-CUDA does.

And bigger picture, on semiconductors it’s the geopolitics people who are watching the bright line: semiconductors are a major power arena, the markets will get adjusted along those lines.

paulmd• 2 years ago

In die area, in TDP, in wafer consumption, it’s just not a big cost.

And yet we live in a world where an AI 370 processor has less than 1/4 of the single-threaded perf/w of a “totally equivalent” M3 Max processor. One of those things can’t be true.

Who are you going to believe, Jim Keller or your lying eyes?

And it only gets worse when you realize that he didn’t vaguely say the thing people claim he said… people have spent 4 years now arguing against observable reality using a quotation from Keller that he didn’t even make lol.

benreesman4• 2 years ago

I find it amusing that it was the off-the-scales "SoC" people at Apple who pioneered the new best paradigm for vector compute. Not surprising mind you, the PA Semi people are terrifying, Apple is one of the best microarchitecture firms in history.

I find this amusing because we still call them "GPU"s. NVIDIA's revenue is like 85-90% datacenter IIRC, gaming is like, anti-trust insurance or something.

The "low end" just almost always eats the "high end".

dzonga• 2 years ago

yeah intel's foundry business might not be competitive in the short term. but it's better than nothing. they can focus on other industries that don't need high performance / for low power.

because if you yield the whole thing to tsmc. you won't be able to recover the expertise, which is catastrophic for both intel / USA.

keepamovin11• 2 years ago

agree. You need to get scrappy and back to basics. Create an internal startup that can act as a seed to rebuild their whole foundry business from the ground up. An incubator to nurture latent capacity and future potential.

Capacity building here has to be a long term play.

ghostpepper• 2 years ago

Anyone else notice how many typos are in this article? Does no one proofread anymore?

stonethrowaway• 2 years ago

Lmao @ AMD coming out on top.

laaserboy• 2 years ago

From the article, “William Shockley, one of the inventors of the transistor at Bell Labs”.

William Shockley did not invent the transistor in 1947. He was not on the patent and was in Europe at the time of the invention. John Bardeen and Walter Brattain were on the patent. Shockley shared the Nobel Prize due to work in 1948 on the transistor effect.

Source: https://www.electronicdesign.com/technologies/analog/article...

netfortius• 2 years ago

And the news just keep getting "better" [1]

[1] https://finance.yahoo.com/news/intel-just-gave-plan-laid-213...

Intel has been talking about the monumental shift that its 20A node will bring for years, but we'll never see it in action.

sumo89• 2 years ago

I'm sure all the points are correct but the assessment for ways forward being to just use TSMC as the manufacturer completely ignores the somewhat perilous position Taiwan is in with a looming Chinese invasion. I'd be very reluctant to tie the future of my business to the fates of a company that may become state property within a year or two.

Intel Honesty

Related Stories