Can Intel Become the Chip Champion the US Needs? (ft.com) 48
Once the leading player in the semiconductor industry, the company is attempting to pull off one of tech's most complex turnrounds. From a report: It was nearly a decade ago when Intel, then the undisputed leader in global semiconductor manufacturing, made a fateful decision. A new technology, extreme lithography, was offering a way to pack more computing power on to the silicon wafers from which tiny chips, essential for widely used products like smartphones and PCs, are cut. Using light to etch complicated integrated circuits, EUV promised an unparalleled degree of miniaturisation, but Intel executives believed it would take years for the method to become practical. Instead, they stuck with older manufacturing techniques for their next generation of chips. This turned out to be a historic mistake, one with consequences that are being felt at a time when the US has put advanced chipmaking at the centre of its national industrial policy. Taiwan Semiconductor Manufacturing Company, which adopted EUV in 2019, has leapfrogged Intel to become the world's most advanced chip manufacturer, closely followed by Samsung. Along with other slips, the judgment call has left Intel -- and the US -- scrambling to catch up.
"Hindsight is 20/20," says Ann Kelleher, head of technology development at Intel and the executive charged with restoring the US chipmaker's manufacturing processes. "It's very easy to look back and say, 'If something different was done...'" Intel is today at another crucial juncture. If, as planned, the company finally produces chips made with EUV in large volume later this year, it will be an important step on the road back. Nowhere will progress be watched more anxiously than in Washington, where the Biden administration is facing an imminent decision about how much financial backing to throw behind the company. Last year's US Chips Act committed $52bn in direct subsidies to support semiconductor manufacturing and boost research and development, along with an estimated $24bn worth of tax credits over the next eight years. The law was designed to reverse a slide that has taken the US share of chip production to 12 per cent, from 37 per cent in 1990. The centrepiece of that plan is to bring leading-edge manufacturing back to the US. For better or worse, that leaves Washington with little choice but to bet heavily on Intel, despite it being the laggard in one of the tech world's most important races. Yet falling behind in advanced chip production is not the only problem hanging over Intel.
Big shifts in its customers' needs -- such as the rise of artificial intelligence -- are threatening to sideline its traditional PC and server chips. Its attempt to go into direct competition with TSMC by becoming a so-called chip foundry, manufacturing chips on behalf of other companies, represents the biggest change to its business since it abandoned its original memory chips for processors nearly 40 years ago. To make things even harder, a yawning financial hole has opened up under the company at just the moment it is trying to make up for years of under-investment with a surge in capital spending. The depth of the reversal, which the company says is caused by a temporary inventory correction, shocked Wall Street in January, when Intel warned its revenue would tumble 40 per cent in the first three months of this year. The setbacks mean that a central piece of US industrial policy is now riding on one of the most difficult and complex tech turnrounds ever attempted. As the US Department of Commerce begins to weigh how to distribute the Chips Act subsidies, deciding how fiercely to back Intel will be a central question.:
"Hindsight is 20/20," says Ann Kelleher, head of technology development at Intel and the executive charged with restoring the US chipmaker's manufacturing processes. "It's very easy to look back and say, 'If something different was done...'" Intel is today at another crucial juncture. If, as planned, the company finally produces chips made with EUV in large volume later this year, it will be an important step on the road back. Nowhere will progress be watched more anxiously than in Washington, where the Biden administration is facing an imminent decision about how much financial backing to throw behind the company. Last year's US Chips Act committed $52bn in direct subsidies to support semiconductor manufacturing and boost research and development, along with an estimated $24bn worth of tax credits over the next eight years. The law was designed to reverse a slide that has taken the US share of chip production to 12 per cent, from 37 per cent in 1990. The centrepiece of that plan is to bring leading-edge manufacturing back to the US. For better or worse, that leaves Washington with little choice but to bet heavily on Intel, despite it being the laggard in one of the tech world's most important races. Yet falling behind in advanced chip production is not the only problem hanging over Intel.
Big shifts in its customers' needs -- such as the rise of artificial intelligence -- are threatening to sideline its traditional PC and server chips. Its attempt to go into direct competition with TSMC by becoming a so-called chip foundry, manufacturing chips on behalf of other companies, represents the biggest change to its business since it abandoned its original memory chips for processors nearly 40 years ago. To make things even harder, a yawning financial hole has opened up under the company at just the moment it is trying to make up for years of under-investment with a surge in capital spending. The depth of the reversal, which the company says is caused by a temporary inventory correction, shocked Wall Street in January, when Intel warned its revenue would tumble 40 per cent in the first three months of this year. The setbacks mean that a central piece of US industrial policy is now riding on one of the most difficult and complex tech turnrounds ever attempted. As the US Department of Commerce begins to weigh how to distribute the Chips Act subsidies, deciding how fiercely to back Intel will be a central question.:
they could (Score:2)
but they won't, they'll screw it up.
Re:they could [except for?] (Score:1)
Not a bad FP, but I think you should have offered at least a hint of your thoughts about why not.
My initial hypothesis is short-term thinking driven by the usual culprit: "Need more profit NEXT quarter." That's mostly based on the experiences of a friend's wife who worked there, but plenty of public evidence, too.
I also think there's a loss of vision problem, too. Founders usually have it, but all of Intel's founders are dead.
Industrial policy and incentives matter, too. Unfortunately those categories seem
Re:they could [except for?] (Score:4)
you should have offered at least a hint of your thoughts about why not.
"Full stack" chip companies are not doing well in the modern world.
The sensible path for Intel is to split into two companies:
1. A fabless chip design company like AMD and Nvidia. Then Fabless-Intel can buy manufacturing services from world-class companies such as TSMC.
2. A chip fab company that can offer services to everyone.
Under the current "full stack" model, Intel can only do as well as the weakest link. The best design in the world won't be the best chip if the fab is substandard. The best fab in the world won't be profitable if it's sitting idle because the design is delayed.
Re: (Score:1)
Interesting thoughts, though (1) I doubt you speak for the author of FP (for whom my comment was intended), (2) I somewhat disagree with your reasoning (in spite of your appeal to ye olde "comparative advantage" (which I do regard as the soundest "result" in the last few centuries of economics as the most dismal "science")), and (3) I'm losing my final hopes for Slashdot as a "bastion" of intelligent discussion (largely due to the collapse of the immoderate moderation system).
Looking over the scare quotes a
Re: (Score:2)
Intel selling off their NAND division was a mistake. Whatever the real reasons, it wasn't about the cash. Those are fantastic products, but were underfunded with certain GTM strategy mistakes. Let's hope that SK doesn't tank Solidigm.
Re: (Score:2)
but they won't, they'll screw it up.
With enough taxpayer money, anything is possible.
Re: (Score:2)
Hahaha, nope. The money will still be gone though.
Commissioner Gordon says... (Score:4, Funny)
So much fab capacity was wasted (Score:2)
Re: (Score:2)
Due to all the taxpayer subsidies being handed out, we are heading into a massive glut of fab capacity, so it doesn't matter if some of it is wasted.
Microchip in my molar (Score:2)
At what point is there diminishing returns in ever-higher circuit density.
Will I get the equivalent of a Cray Y-MP in my molar? If I do, what purpose will it serve?
Re: (Score:2)
At what point is there diminishing returns in ever-higher circuit density?
We've already hit that point on cost. Transistors are still getting smaller, but they are no longer getting cheaper.
Will I get the equivalent of a Cray Y-MP in my molar?
A Cray Y-MP was 400 MIPS in 1988.
Today, you can buy a wristwatch with far more processing power than that.
Go AMD (Score:2, Insightful)
Too many suits at Intel. We need at least 2 players in the chip business. More would be better.
Re: (Score:2)
Too many suits at Intel. We need at least 2 players in the chip business. More would be better.
Two players? We need another early 90's situation where there were up to a dozen major companies making clone X86 chips. Some under license, some independently engineered. Most of them ended up getting bought out by the big two, but at one time, everyone from Texas Instruments to NEC was making X86 chips. That's not to mention all the companies that have since been absorbed, like Cyrix and IDT. It'd be nice to have that kind of variety in X86 again.
Re: (Score:2)
Re:Go AMD (Score:4, Informative)
That would be a huge change given that AMD hasn't been in the chip manufacturing business since 2009.
Re: (Score:2)
Too many suits at Intel. We need at least 2 players in the chip business. More would be better.
The absolute minimum number of companies required for competitive behavior to emerge is 4, and that only applies in sectors that don't involve natural monopolies. The FTC should be aggressively preventing mergers and acquisitions that reduce the count of major competitors in any sector below 4.
Unfortunately it's a little late practically everywhere.
Focused on the wrong thing (Score:3)
All china has to do is tell the local chinese manufacturerrs that they must use only locally made chips and then put an import tariff on MiA. That would kill off this 52B.
OTOH, If simply 2B was pulled out of that money and used to get new Ideas started, along with getting board manufacturing increased here in the states, then it becomes what is needed.
But America has made numerous starts on various areas, without realizing that what is needed is to have all of the line from mine back to final product, OR have points in there where we get parts, elements, etc from friendly/neutral nations, while still bringing back the rest.
ICF round 2? (Score:5, Interesting)
I interviewed for a job in the "Intel Custom Foundry" group when I was finishing my masters degree. It seemed like a cool idea ~10 years ago to sell Intel's manufacturing capacity to other companies, but it ultimately didn't work out and shut down 5 years ago, attracting mostly just FPGA companies.
I'm curious why Intel Custom Foundry didn't work, but why Intel Foundry Services will work now?
Re: (Score:2)
It only affected the smaller ones. Intel bought the biggest FPGA company that was using Intel's foundry services, Altera.
One reason that Intel's foundry services didn't succeed is that companies were reluctant to become dependent on a company that was also a direct competitor. That would still be a problem for companies looking to make desktop or laptop CPUs; if such a company appears it's likely to turn to TSMC to make its chips, as Apple and Qualcomm do. Right now those are the only two companies other th
Honestly thought we were past this naivite (Score:3)
Ignormus journalists (Score:3)
"Using light to etch complicated integrated circuits, EUV promised ..."
Idiots. Light has been used to expose photoresist prior to etching since the 1960s. As I understand it what is new in EUV is simply the wavelength, which thus allows smaller feature size, but at the expense of a very difficult process.
Re: (Score:2)
EUV requires the optical path to be a vacuum and mirrors for focusing rather than lenses. Since even the mirrors absorb about a third of the incident EUV each the source must be significantly brighter.
EUV sources are not lasers, so the beam is not monochromatic and focusing is subject to chromatic aberration.
Intel had - and has - lots of problems (Score:4, Insightful)
Their choice not to pursue EUV back then is unrelated to most of them. But I can see why Intel would try to steer the discussion that way... much easier to pretend everything is being fixed now.
They donâ(TM)t have to be (Score:3)
Re: (Score:1)
Re: (Score:3)
I would not count on it. Nobody thought the West would deliberately sabotage the energy supply to central Europe over the Ukraine conflict either; and now it sure as a hell looks like some combination of the Biden admin and the US deep state did example that...
I strongly suspect if it became apparent there was real risk of the ROC capitulating to PRC forces, US intelligence would ensure there was an 'accident' that would take TSMC's assets out of production for at least some years.
Re: (Score:2)
> TSMC will not be destroyed, just under new ownership.
Why would you let your enemies take your assets? Better to destroy them then let the enemy benefit. War 101.
Re: (Score:3)
If TSMC is destroyed Intel will suddenly be the world leading FAB, which means even if TSMC could have done better, they are gone now so Intel is the only game in town
TSMC will not be destroyed, just under new ownership.
It's highly likely that Taiwan suffers one of two fates in the next decade: Invasion and military conquest by China, or a relatively peaceful process by which China essentially absorbs the island and its people through political and economic means. Unless the US commits to nuclear exchange to prevent it... highly unlikely... the smart bet is that China will through either method, eat Taiwan in a Borg-like manner.
So in either case, TSMC will either be destroyed outright in war, or belong to the Chinese Commu
Re: (Score:2)
So you are saying it's in the US's interest to see a war over Taiwan.
By amazing coincidence US media reports of Chinese aggression over Taiwan have increased recently.
Re: (Score:2)
It's deeper than chips.
BRICS are abandoning the petrodollar. No one is going to be buying the growing US debt.
The only way to get out from under China's thumb is to beat them in a war and abolish the US debt as punishment for losing.
Re: (Score:2)
Headline could be... (Score:1)
Can Intel ruin Taiwan's chip sector like what happened to Japan, after US chipmakers in the early 90s copied Japan's manufacturing techniques, subsidized, etc.
Strategy time (Score:5, Funny)
Intel needs to go back to its original playbook.
When times get tough, dream up yet another unorthodox byzantine processor architecture based on the latest industry buzzwords. Then wait a few years for it to fail in the marketplace due to abysmal performance. Finally, set up a skunkworks project to further improve CPUs derived from the original 8008 to get impressive performance improvements, and sell these for fun and profit.
Re: (Score:2)
Itanic 2 - The revenge of Intel.
Re: (Score:2)
The original playbook of Intel is making memory. They never really understood CPUs and it still shows.
Re: (Score:2)
Comment removed (Score:3)
Re: (Score:2)
Last I checked, SR SKUs aren't in Passmark yet.
And FWIW a benchmark may or may not reflect performance on a given workload, especially without eldritch NUMA and BIOS tweaks.
Betteridge has you covered (Score:2)
If you have to ask, the answer is No
Does not matter for most customers (Score:2)
Intel has tried a few times to be a fab for those outside of Intel. It failed.
Even inside Intel they use TSMC and other fabs for a lot of non-CPU work because the internal fab is completely interwoven with the CPU side of things. The DRC/LVS rules are impenetrable to say the least.
As an example a local university my work was affiliated was given a free shuttle run at Intel. The group of grad students and advisor was familiar with TSMC with multiple successful tape-outs on comparable nodes. After weeks o
old guys, still there (Score:2)
The army of middle managers at Intel, the ones that have been there forever
because they're mostly good at dodging getting laid off, are all praying for a day when
their pile of stock vests at 1990's prices. They all discuss it at lunch, every day.
Those guys ain't changing nuthin'.
No (Score:2)
Intel can't even produce a chip that isn't a buggy piece of shit.