News U.S. Govt pushes Nvidia and Apple to use Intel's foundries — Department of Commerce Secretary Raimondo makes appeal for US-based chip production

[Dustyboy1492]

"while Raimondo did not mention Intel during the meetings"

Sounds like this entire article is speculative and misleading.

 

[phead128]

No amount of free tax payer money and xenophobic fearmongering can overcome Intel's deep mismanagement and leadership culture. That's just the unfortunate fact.

 

[MacZ24]

No amount of free tax payer money and xenophobic fearmongering can overcome Intel's deep mismanagement and leadership culture. That's just the unfortunate fact.

Pretty much the problem of the US/west right now ... trying to fight cancer by engaging in fights with the neighbors.

Even if you win (you won't), you're still dying from cancer.

Like (almost) every empire before it, the US is trying to stop its (internal) decline by waging (external) conflicts.

And none of the geniuses that rule the show seems to understand that they killed the golden goose and just seem to be able to double down on the toxic behavior.

 

[Notton]

Broadcom: Intel sucks
Apple: You know we ditched Intel a few years ago, right?
AMD: Surely you must be joking.
Nvidia: Why would we use Intel?
Intel: TSMC ftw

also Intel: Why is no one lining up to order from us?

 

[Dntknwitall]

Then add tariffs on foreign subsidized chips. Taiwanese, Korean, Japanese, and of course Chinese governments have been subsidizing their domestic chip production for decades. It is time for the US to match their subsidies with CHIPS act and then add tariffs on foreign subsidized chips.

You know what that would do? You have no idea what you are talking about. Tariffs on TSMC would crash the tech market in the US and Canada. Prices would need to go up and sales would plummet. Therefore crashing all stocks and destroying the global market. Glad you have nothing to do with politics.

 

[watzupken]

Every government should push its manufacturers to use its own nation's goods. Not sure why everyone got wedgies.

This is called protectionism. It may sound logical, until you figured you are paying more than double for the same good manufactured in different location. In the end, people will just import the cheaper version and no help to consumer and hiring.

 

[renz496]

Broadcom: Intel sucks
Apple: You know we ditched Intel a few years ago, right?
AMD: Surely you must be joking.
Nvidia: Why would we use Intel?
Intel: TSMC ftw

also Intel: Why is no one lining up to order from us?

Isn't that nvidia already confirmed to use intel? Also about a decade ago Jensen publically said Intel should open their fab. But back then intel just smirk at Jensen and think their fab prowess will go on forever. And they want to keep it to themselves only.

 

[Gururu]

This is called protectionism. It may sound logical, until you figured you are paying more than double for the same good manufactured in different location. In the end, people will just import the cheaper version and no help to consumer and hiring.

I buy car parts from non-American companies but I buy my CPU from American company. It’s because I buy from the best. I would probably still do so if the prices doubled so long as they remained the best.

 

[steve15180]

AMD is the one that reverse engineered Intel's chips when they were fabbing them, then made and sold them as AMD chips.

Intel has never done anything like this. AMD has patents on their designs. The components themselves I.E transistors, sram, etc are off the shelf bits. Intel is not allowed to just violate their patents and steal their designs.

Where did you get that from? AMD reverse engineered NOTHING. Compaq reverse IBM's bios, but both AMD and Cyrix have legitimate X86 licenses granted by Intel. Which, by the way, Intel spent a good deal of time trying to find a way to not honor the license until the courts said otherwise. In order to get the IBM PC business, IBM insisted on second sourcing the original chips because Intel could not supply everything IBM needed.

 

[steve15180]

AMD did indeed start out manufacturing licensed Intel products. That was when they still had fabs of their own and volume was needed as the x86 market was exploding.

AMD did reverse engineer Intel's early x86 processors, but they didn't actually steal anything. They just made a 286 compatible processor and launched as a competitor. Basically, they looked at how it needed to be done, and designed their own silicon to do it. Any overlap was simply the result of them being made to the same purpose. They weren't perfect, and in many cases over the next several generations weren't as performative as Intel chips.

They also did similar with MMX and other Intel innovations so they could maintain that compatibility. AVX would be another, more recent example where AMD had to follow suite.

Some lawsuits did happen when infringement was perceived and they basically agreed to cross license x86 for amd64. Intel can't break with AMD and AMD can't break with Intel or neither can make useful x64 chips any longer.

Intel licensed Nvidia and AMD GPU tech to make their integrated graphics as well. And Intel is just now getting away from that with Xe cores.

I want to add that we also had Cyrix and Via making consumer x86 products back then.

AMD reverse engineered nothing, Intel granted a license to manufacture x86 to AMD and Cyrix because, at that time, Intel could not supply IBM with the chips they needed on their own. Eventually, AMD engineers did start tinkering with ways to improve on the original chips on there own while Intel did their thing. Hence, x86 is about all they have in common anymore.

It was sometime later when the court slapped down Intel for trying to make x86 proprietary again, and the ability to cross license critical parts of x86 came to be.

Intel actually rejected AMD 64, as they, again, wanted the whole pie to themselves and launched Itanic as the 64 bit future for everyone. Needless to say, the Itanic sunk. As a result, IA64 was developed, a close enough for government clone of AMD64 for their processors.

 

[Amdlova]

American gov is right. Need to take some of these tech and place on american company... Taiwan will be chinese again. It's a matter of time. Asia war 2030.
The Allies will sit and watch.

 

[Mikhail Ratchkovski]

AMD did indeed start out manufacturing licensed Intel products. That was when they still had fabs of their own and volume was needed as the x86 market was exploding.

AMD did reverse engineer Intel's early x86 processors, but they didn't actually steal anything. They just made a 286 compatible processor and launched as a competitor. Basically, they looked at how it needed to be done, and designed their own silicon to do it. Any overlap was simply the result of them being made to the same purpose. They weren't perfect, and in many cases over the next several generations weren't as performative as Intel chips.

They also did similar with MMX and other Intel innovations so they could maintain that compatibility. AVX would be another, more recent example where AMD had to follow suite.

Some lawsuits did happen when infringement was perceived and they basically agreed to cross license x86 for amd64. Intel can't break with AMD and AMD can't break with Intel or neither can make useful x64 chips any longer.

Intel licensed Nvidia and AMD GPU tech to make their integrated graphics as well. And Intel is just now getting away from that with Xe cores.

I want to add that we also had Cyrix and Via making consumer x86 products back then.

They did not steal it, they just looked at how it needed to be done. LOL that is quite funny!

 

[darkchanter]

sounds like planned economy to me, China-style

 

[stuff and nonesense]

AMD is the one that reverse engineered Intel's chips when they were fabbing them, then made and sold them as AMD chips.

Intel has never done anything like this. AMD has patents on their designs. The components themselves I.E transistors, sram, etc are off the shelf bits. Intel is not allowed to just violate their patents and steal their designs.

Amd is the company that was the second source for x86. IBM stipulated that a second source was required, my Dell 200 had a chip from AMD labelled as copyright Intel 1982.

Intel broke the licensing agreement with the 386 (and lost in court). The court case put AMD a generation behind as they were effectively prohibited from manufacturing and selling 386 chips for the duration of the litigation. The I486 was released around the time the court case finished.

The Am486 was a reverse engineered clone, the Pentium equivalent was a code compatible chip and AMD x86 chips since have been code compatible.

Don’t forget the Intel of the 2000-2010 era, the intel that operated with bungs and anticompetitive practices to freeze out viable competition.. there are no angels in business.

 

[nogaard777]

Yeah, imagine Apple's next chip fabbed by Intel. lol

Considering Intel is about to leapfrog TSMC and Apple is more than happy to pay for cutting edge nodes I can not only imagine it, but I almost expect it.

 

[nogaard777]

The problem with US based foundries is that they rarely have the scale available. This isn't going to change in the near term and I'd be surprised if leading edge volume capacity was ready before the end of 2026. The financial bros running Intel gambled on the company's future by not adopting EUV despite being one of the companies investing in getting it done.

The one place Intel does have a distinct advantage right now is in packaging, but I doubt that's very lucrative without the fabrication to go along with it. They'd have to convince potential customers to use their packaging technology and design around it rather than TSMC's.

Huh? Intel is the FIRST to implement EUV, not TSMC. What do you think 20A and 18A are on? TSMC specifically pulled back from EUV due to its high costs and they're already making so much profit on their 3,4, and 5nm nodes it doesn't make financial sense when they can milk the contracts they're already sitting on.

Intel is ahead of TSMC, not behind. TSMC isn't scheduled to start mass production of 2nm until late '25, while Intel was already on track to start this fall, but cancelled it since 18A is so close behind it will likely come online before TSMC's 2nm.

You also have to keep in mind TSMC keeps raising prices so wafers that were $5-$7k just a few short years ago are now topping $20k+. I have little doubt 3rd parties will eventually get tired of TSMC's pricing as well as pitting customers against each other to compete for wafer allocation.

 

[nogaard777]

“Intel CEO Pat Gelsinger expressed frustration with America’s reliance on TSMC to produce advanced chips”

Pat forgot he is also using TSMC for his chips. Who is he frustrated with? Himself? So if yield is poor, why would companies use Intel foundry.

I always see this mindset when Intel can't magically bring new nodes online instantly. 18A is very much on schedule and producing acceptable yields, but you need to understand the scale of production Intel needs to shift it's entire product lineup to a new node. Intel is only using TSMC for a portion of Alder Lake, not the entire package. TSMC simply CAN'T provide enough silicon for Intel's needs, even if they wanted to. They would have to tell off Apple, Qualcomm, AMD, Nvidia, etc. For every DIYer or enthusiast fanboying for AMD's X3D chips, Intel makes dozens of CPUs for the likes of Dell, HP, and Lenovo.

Then you have to keep in mind while TSMC might technically produce more silicon than Intel, half of that is on legacy nodes for things that aren't CPUs and GPUs. But the majority of Intel's silicon has to be cutting edge or near cutting edge.

Much like with 10nm I expect they'll first shift laptop lines to the newest node as laptops benefit the most due to efficiency, then as 18A ramps up the desktop lines will shift as well.

 

[nogaard777]

At least this article noticed that Intel has yet to receive payment of anything from that Chips act from around 3 years ago. But because of misinformation spreading journalists like this author it is common knowledge that Intel has already received tens of billions of dollars.

Also "common knowledge" are the misconceptions that all Intel 13th and 14th gen chips are failing, low volt instability=high volt degradation, and Intel CPUs generally use 320w all of the time.

This article is trying to spread another one: "Because of this, Intel is reportedly considering spinning off its manufacturing " is per one possibly short selling analyst and should not be portrayed as mainstream fact as it is not.

The bias held by the author in some of these articles apparently completely justifies spreading misinformation. Are you more or less informed after reading them?

I've definitely noticed an "Intel bad" bias growing within Tom's. AMD has always had the loudest fanboys on tech sites, Youtube, and Reddit, and more than a few tech media have pointed it out that if you want clicks you just follow that "AMD good, Intel bad" line. I've been doing this for over 25 years so I hate all these companies, but while AMD has always had the loudest fanboys, I've noticed more and more AMD bias from the media the last several years that wasn't there before, and when I see straight up misinformation like you listed above it's easy to see why I almost never read Tom's anymore. Every time I get a link to an article I may as well be watching Moore's Law Is Dead because I know anything about Intel is going to be doom and gloom while AMD is going to be defended. Just look at Tom's Zen5 review. Literally the most cherry picked benchmarks to show Zen5 in it's most positive light that no one else in tech media saw.

 

[rluker5]

Amd is the company that was the second source for x86. IBM stipulated that a second source was required, my Dell 200 had a chip from AMD labelled as copyright Intel 1982.

Intel broke the licensing agreement with the 386 (and lost in court). The court case put AMD a generation behind as they were effectively prohibited from manufacturing and selling 386 chips for the duration of the litigation. The I486 was released around the time the court case finished.

The Am486 was a reverse engineered clone, the Pentium equivalent was a code compatible chip and AMD x86 chips since have been code compatible.

Don’t forget the Intel of the 2000-2010 era, the intel that operated with bungs and anticompetitive practices to freeze out viable competition.. there are no angels in business.

While what you say may be correct, it is omitting what I said so it is only half truth: https://en.wikipedia.org/wiki/AMD_Am9080
AMD did not have a prior agreement and did not respect IP rights with their first microprocessor when they released it.

Also Intel breaking the licensing 286 agreement? While AMD may have won in court,:

"While the AM386 CPU was essentially ready to be released prior to 1991, Intel kept it tied up in court.[2] Intel learned of the Am386 when both companies hired employees with the same name who coincidentally stayed at the same hotel, which accidentally forwarded a package for AMD to Intel's employee.[3] AMD had previously been a second-source manufacturer of Intel's Intel 8086, Intel 80186 and Intel 80286 designs, and AMD's interpretation of the contract, made up in 1982, was that it covered all derivatives of them. Intel, however, claimed that the contract only covered the 80286 and prior processors and forbade AMD the right to manufacture 80386 CPUs in 1987. After a few years in the courtrooms, AMD finally won the case and the right to sell their Am386 in March 1991.[4] This also paved the way for competition in the 80386-compatible 32-bit CPU market and so lowered the cost of owning a PC.[1]" from: https://en.wikipedia.org/wiki/Am386

This does not portray how second sourcing works. You do not second source behind the back of the source. Once again an example of AMD not respecting IP rights. If the 386 is a 286, why isn't Lunar Lake? 286 is 16 bit, 386 is 32 bit. Totally sleazy.

My point was
"AMD is the one that reverse engineered Intel's chips when they were fabbing them, then made and sold them as AMD chips.

Intel has never done anything like this. AMD has patents on their designs. The components themselves I.E transistors, sram, etc are off the shelf bits. Intel is not allowed to just violate their patents and steal their designs."

Complaining about Intel's dealings with Dell is separate to this and Intel's respect for other's IP in regards to the foundry industry.

 

[rluker5]

Huh? Intel is the FIRST to implement EUV, not TSMC. What do you think 20A and 18A are on? TSMC specifically pulled back from EUV due to its high costs and they're already making so much profit on their 3,4, and 5nm nodes it doesn't make financial sense when they can milk the contracts they're already sitting on.

Intel is ahead of TSMC, not behind. TSMC isn't scheduled to start mass production of 2nm until late '25, while Intel was already on track to start this fall, but cancelled it since 18A is so close behind it will likely come online before TSMC's 2nm.

You also have to keep in mind TSMC keeps raising prices so wafers that were $5-$7k just a few short years ago are now topping $20k+. I have little doubt 3rd parties will eventually get tired of TSMC's pricing as well as pitting customers against each other to compete for wafer allocation.

You are thinking of high NA EUV that Intel is first in. TSMC has been making lower NA EUV for a few years now.

 

[stuff and nonesense]

My point was
"AMD is the one that reverse engineered Intel's chips when they were fabbing them, then made and sold them as AMD chips

From Tom’s
“In 1985, Intel released its first 32-bit x86 processor design, the 80386. AMD planned to release its variation, the AM386, not long after, but Intel held it up in court. Intel claimed that its cross-licensing agreement permitted AMD to produce copies of only the 80286 and older processor designs, but AMD argued that the contract permitted it to create clones of the 80386 and future x86 derivatives, as well. After years of legal battles, the courts sided with AMD, and the company was able to release its AM386 in 1991.”

Nowhere in your reference does it say that AMD was fabbing i386 CPUs.

The Court agreed with AMD, also AMD produced 2 486 variants, one with Intel microcode, one wholly AMD, from there forward the chips have been code compatible and a patent cross license arrangement exists between AMD and Intel.

Intel has never done anything like this

How about not respecting the patent that describes the finfet manufacturing process on which most of the core architecture has been built?
https://www.tomshardware.com/news/i...challenge-against-chinese-academy-of-sciences

CN102956457B - Semiconductor device structure, manufacturing method thereof and semiconductor fin manufacturing method - Google Patents

The application discloses a semiconductor device structure, a manufacturing method thereof and a semiconductor fin manufacturing method. The manufacturing method of the semiconductor device structure comprises the following steps: providing a semiconductor substrate; forming a fin on a...

patents.google.com

 

[Notton]

Isn't that nvidia already confirmed to use intel? Also about a decade ago Jensen publically said Intel should open their fab. But back then intel just smirk at Jensen and think their fab prowess will go on forever. And they want to keep it to themselves only.

I thought that was a rumor for an ARM processor?
If true, I am sure Nvidia got a killer deal on it, because I would have expected them to fab it at Samsung.

 

[rluker5]

From Tom’s
“In 1985, Intel released its first 32-bit x86 processor design, the 80386. AMD planned to release its variation, the AM386, not long after, but Intel held it up in court. Intel claimed that its cross-licensing agreement permitted AMD to produce copies of only the 80286 and older processor designs, but AMD argued that the contract permitted it to create clones of the 80386 and future x86 derivatives, as well. After years of legal battles, the courts sided with AMD, and the company was able to release its AM386 in 1991.”

Nowhere in your reference does it say that AMD was fabbing i386 CPUs.

1. You aren't trying to refute that AMD made an unlicensed, reverse engineered clone of the Intel 8080 so I'll take that as an admission on your part that AMD did in fact be untrustworthy in respect to a fab customer's IP.
2. You are the one who brought up the 286 licensing agreement and it is not disputed that AMD cloned the 386 without the knowledge of Intel and planned to sell it as a "second source" against Intel's intentions in signing a second source agreement for an earlier chip that was about as different as AM64 chips. I responded because it was another good example of AMD not respecting IP, not because it was exactly like the 8080 example.

The Court agreed with AMD, also AMD produced 2 486 variants, one with Intel microcode, one wholly AMD, from there forward the chips have been code compatible and a patent cross license arrangement exists between AMD and Intel.


How about not respecting the patent that describes the finfet manufacturing process on which most of the core architecture has been built?
https://www.tomshardware.com/news/i...challenge-against-chinese-academy-of-sciences

CN102956457B - Semiconductor device structure, manufacturing method thereof and semiconductor fin manufacturing method - Google Patents

The application discloses a semiconductor device structure, a manufacturing method thereof and a semiconductor fin manufacturing method. The manufacturing method of the semiconductor device structure comprises the following steps: providing a semiconductor substrate; forming a fin on a...

patents.google.com

Intel was in volume production of finfet chips less than 2 months after that patent was filed. And Intel had published information on how they work months before. Intel called it Tri-Gate at the time.
Here's some information on the timeline:
Chinese patent filed: https://patents.google.com/patent/CN102956457B/en
Tom's article on Ivy Bridge: https://www.tomshardware.com/news/ivy-bridge-processor-release-22nm-3d-transistor,13753.html
Intel's press release on Finfet prior to the Chinese patent: https://download.intel.com/newsroom/kits/22nm/pdfs/22nm-Details_Presentation.pdf

It still looks like Intel is the party that respects intellectual property rights.

 

[thestryker]

Huh? Intel is the FIRST to implement EUV, not TSMC. What do you think 20A and 18A are on? TSMC specifically pulled back from EUV due to its high costs and they're already making so much profit on their 3,4, and 5nm nodes it doesn't make financial sense when they can milk the contracts they're already sitting on.

No Intel missed the boat on EUV as TSMC's first EUV node was N7 which debuted consumer parts in 2018. TSMC and Samsung both have more EUV machines than Intel which allows them higher capacity. This lack of machines has cost Intel a fair bit of money already and is a plausible reason behind them abandoning 20A for 18A.

You're probably thinking of High-NA EUV machines, but those aren't being used in fabrication yet and aren't scheduled to until 14A. The first test machine started being installed at the end of last year.

Intel is ahead of TSMC, not behind. TSMC isn't scheduled to start mass production of 2nm until late '25, while Intel was already on track to start this fall, but cancelled it since 18A is so close behind it will likely come online before TSMC's 2nm.

Intel is not ahead of TSMC on production nodes. Intel has GAAFET and BSPDN coming with 18A (unless there's still a line using 20A, but that doesn't seem likely) which are things that TSMC has yet to implement. Once 18A goes high volume it ought to be the most advanced node available if they hit their time tables.

You also have to keep in mind TSMC keeps raising prices so wafers that were $5-$7k just a few short years ago are now topping $20k+. I have little doubt 3rd parties will eventually get tired of TSMC's pricing as well as pitting customers against each other to compete for wafer allocation.

TSMC only gets away with this because there's no alternative. That's the biggest advantage Intel has for breaking into the leading edge market. The biggest problem Intel has is scale due to not having enough EUV capacity. This is the reason I hope is why they moved on from 20A to just go 18A. This would allow them to avoid another MTL situation where they have to keep a one off node going to meet manufacturing requirements instead of using those machines for a long term node.

 

[rtoaht]

That can include Intel chips fabbed by TSMC.

Sure. That should give Intel more impetus to execute on 18A and bring their client chips in-house just like their data center chips which are already made in-house.