News Intel's CEO Fires Back at 3nm Delay Rumors

[Admin]

Pat Gelsinger: Both internal and external 3 nm projects are on track.

Intel's CEO Fires Back at 3nm Delay Rumors : Read more

 

[rluker5]

What are these rumors?
Is this more MLD stuff?

 

[Deleted member 2731765]

What are these rumors?
Is this more MLD stuff?

You mean Moore's Law is Dead YT channel stuff/MLID rumors ? Or, some other thing ? Nope, these are not rumors by the way. Not by MLID though.

It's a response/confirmation from Pat Gelsinger, chief executive of Intel, that they are on track with 3nm node, because recently DIGITIMES leaked out this rumor citing industry sources that Intel might delay it's 3nm orders, which was NOT true to begin with.

 

[MBOO7]

No offense, but trusting Pat is like trusting a liar. Look who cut the dividend suddenly (not to mention all those insider sales shortly before the announcement).
Even (crappy) WCC has a better reputation than Pat.

 

[thisisaname]

Released in 2024? would that be the same kind of release the ARC GPU got when it first came out. A very quite and limited number somewhere in China?

 

[Giroro]

I think this is just all a bunch of CEO weaseling, since "Intel 3" is a deliberately-misleading rebrand of Intel's "5 nm class" process. Its performance/efficiency will be competitive with TSMC N5. "Intel 3" could be delayed or even cancelled without affecting "Intel's 3nm" and vice versa.

Although I don't think Gelsinger actually cares what anybody thinks in regards to delays, just as long as he can continue tricking people into conflating Intel's generation-behind tech with TSMC's superior bleeding-edge.

 

[jamok99]

Gelsinger is such an unabashed liar, he probably wears asbestos pants to keep from getting continually singed. Intel hasn't been on time with many, if not most, of their important products since 2016, due to constant delays with perfecting their die shrinks. You know a company's in trouble when current promises turn to dust and the CEO focuses on future promises, which turn to dust when you arrive there.
In particular, ""Good solid execution on both the client, the server, and AXG [accelerated graphics] side. We are gaining momentum with foundry customers as well. So, I feel good that we have turned the corner on many of the execution challenges. You know, these rumors, like many others, will be proven by our execution to be firmly false."
I recently read that Intel's program (can't recall the name of it) to lure customers to its own fabs (in an attempt to compete with TSMC), has resulted in exactly 1 contract to date. So that's what "gaining momentum with foundry customers as well" really means. Sheesh.
If you want to see an immediate 'bump up' in Intel's stock price, fire Gelsinger and replace him with someone who's competence extends beyond just words.

 

[brandonjclark]

No offense, but trusting Pat is like trusting a liar. Look who cut the dividend suddenly (not to mention all those insider sales shortly before the announcement).
Even (crappy) WCC has a better reputation than Pat.

So, you're a big Intel fan, then?

😉

 

[rtoaht]

...Its performance/efficiency will be competitive with TSMC N5...

Any source for this statement other than "just trust me bro"?

 

[rtoaht]

No offense, but trusting Pat is like trusting a liar. Look who cut the dividend suddenly (not to mention all those insider sales shortly before the announcement).
Even (crappy) WCC has a better reputation than Pat.

In the Q4 2022, Intel CFO mentioned that Intel will offer "competitive dividend". They didn't promise they wouldn't reduce it at all. Which part is the lie?

 

[rtoaht]

So, you're a big Intel fan, then?

😉

Very few billion dollars corporations have die-hard fans. There are few notable exceptions, however, like AMD or Tesla.

 

[jkflipflop98]

The thing about this industry is unless you've dedicated your entire life to it, you can't really understand what's going on. The average joe on the street has zero clue how a CPU is made or how long it takes to fab one. They don't know and don't care. So you can make up ANYTHING you like and people will just accept it as fact because there's no way they could know any better.

 

[jkflipflop98]

No offense, but trusting Pat is like trusting a liar. Look who cut the dividend suddenly (not to mention all those insider sales shortly before the announcement).
Even (crappy) WCC has a better reputation than Pat.

Oh you children and your short little memories. Pat is the OG. He is due his respect.

 

[DonQuixoteIII]

Very few billion dollars corporations have die-hard fans. There are few notable exceptions, however, like AMD or Tesla.

You forgot your sarcasm tags,, Or, if you didn't, then your truth-meter is out of adjustment.

 

[SiliconFly]

I think this is just all a bunch of CEO weaseling, since "Intel 3" is a deliberately-misleading rebrand of Intel's "5 nm class" process. Its performance/efficiency will be competitive with TSMC N5. "Intel 3" could be delayed or even cancelled without affecting "Intel's 3nm" and vice versa.

Although I don't think Gelsinger actually cares what anybody thinks in regards to delays, just as long as he can continue tricking people into conflating Intel's generation-behind tech with TSMC's superior bleeding-edge.

Wrong. Many credible reports including toms own article compared the node characteristics and clearly stated that Intel 4 is actually on par with TSMC N3 in many aspects (power, performance, density, etc). And the nextgen Intel 3 actually is not only on par with TSMC N3, but also exceeds it in certain aspects. Read the comparisons.

But this is all old news. Intel 3 is set be an "large" IFS node with a host of PDKs & libraries that'll cater not only to intel, but also to many other customers as well. Boring.

More recent exciting news is, Intel is coming out with it's high performance 20A node next year custom built for Arrow Lake only. And it's far far ahead of TSMC N3. Its a 2nm-class node with gate-all-around FET & backside-power-delivery. The first of it's kind in the industry. Beats TSMC 3nm by a mile.

(Note: TSMC N3 is based on old technology like Intel 3, Intel 4 & TSMC N5).

 

[SiliconFly]

What are these rumors?
Is this more MLD stuff?

MLID youtube channel spreads a lot of fake news about Nvidia & Intel all the time. And worse, he praises AMD even when it crashes & burns! So fake...

 

[waltc3]

For Gelsinger, if you'll notice, when he says, "good, solid, execution" he always is talking future tense. Talking about what you may be able to do at some point in 2024 is not talking about anything that is presently the product of "good, solid, execution." It's like when Gelsinger said he had "put AMD in the rearview mirror" he was obviously driving backwards...!

 

[Kamen Rider Blade]

Gelsinger is such an unabashed liar, he probably wears asbestos pants to keep from getting continually singed. Intel hasn't been on time with many, if not most, of their important products since 2016, due to constant delays with perfecting their die shrinks. You know a company's in trouble when current promises turn to dust and the CEO focuses on future promises, which turn to dust when you arrive there.
In particular, ""Good solid execution on both the client, the server, and AXG [accelerated graphics] side. We are gaining momentum with foundry customers as well. So, I feel good that we have turned the corner on many of the execution challenges. You know, these rumors, like many others, will be proven by our execution to be firmly false."
I recently read that Intel's program (can't recall the name of it) to lure customers to its own fabs (in an attempt to compete with TSMC), has resulted in exactly 1 contract to date. So that's what "gaining momentum with foundry customers as well" really means. Sheesh.
If you want to see an immediate 'bump up' in Intel's stock price, fire Gelsinger and replace him with someone who's competence extends beyond just words.

Intel's Fab won't be gaining much external customers until they seperate Intel (Chip Design) from their foundaries.

Similar to how TSMC has no (Chip Design) team, but just works as a "PURE Foundary" that contracts out to clients.

Until Intel is willing to split up the Foundary side from their (Chip Design) side and be seperate companies with no more ties, like AMD did with Global Foundaries, they'll always be bogged down by the (Chip Design) side and the lack of trust from working with their foundaries and if the foundary side will steal IP and knowledge to help the (Chip Design) side.

It's like we need a seperation of Church & State, but for (Chip Foundary & Chip Design).

You can do one only, you can't be both.

TSMC's entire business model is predecated on that seperation, Intel could stand too learn from TSMC and split themselves into two companies.
It'd be better for everybody around the world if they split up.

That's how you maintain trust.

 

[shady28]

Intel's Fab won't be gaining much external customers until they seperate Intel (Chip Design) from their foundaries.

Similar to how TSMC has no (Chip Design) team, but just works as a "PURE Foundary" that contracts out to clients.

Until Intel is willing to split up the Foundary side from their (Chip Design) side and be seperate companies with no more ties, like AMD did with Global Foundaries, they'll always be bogged down by the (Chip Design) side and the lack of trust from working with their foundaries and if the foundary side will steal IP and knowledge to help the (Chip Design) side.

It's like we need a seperation of Church & State, but for (Chip Foundary & Chip Design).

You can do one only, you can't be both.

TSMC's entire business model is predecated on that seperation, Intel could stand too learn from TSMC and split themselves into two companies.
It'd be better for everybody around the world if they split up.

That's how you maintain trust.

But that's already happening. IFS has contracts with Amazon, Qualcomm, MediaTek, and the DoD - that we know of. Another announcement was for a major cloud vendor - likely either Amazon, Google, or Microsoft. Most of this appears to be on the future Intel 3 mode, while the DoD contract is for the 20A and later 18A nodes.

It will be interesting to see what starts coming out of these new fabs. This should start to show up sometime in 2024.

And lets be real, outside of Apple, nobody has access to TSMC N3.

So the foundry fight starts in 2024.

Apple Orders Entire Supply of TSMC's 3nm Chips for iPhone 15 Pro and M3 Macs

Apple has reportedly secured all available orders for N3, TSMC's first-generation 3-nanometer process that is likely to be used in the upcoming...

www.macrumors.com

 

[thestryker]

Intel's Fab won't be gaining much external customers until they seperate Intel (Chip Design) from their foundaries.

Similar to how TSMC has no (Chip Design) team, but just works as a "PURE Foundary" that contracts out to clients.

Until Intel is willing to split up the Foundary side from their (Chip Design) side and be seperate companies with no more ties, like AMD did with Global Foundaries, they'll always be bogged down by the (Chip Design) side and the lack of trust from working with their foundaries and if the foundary side will steal IP and knowledge to help the (Chip Design) side.

It's like we need a seperation of Church & State, but for (Chip Foundary & Chip Design).

You can do one only, you can't be both.

TSMC's entire business model is predecated on that seperation, Intel could stand too learn from TSMC and split themselves into two companies.
It'd be better for everybody around the world if they split up.

That's how you maintain trust.

This post shows an impressive amount of ignorance as to how chip making/business works. No company looking at an external fab is going to be concerned with anything you're suggesting. Even if Intel were to steal something it would take them years to implement it, they would destroy their reputation and have nasty legal ramifications. The biggest actual concern is lack of access to leading edge nodes and Intel prioritizing its own products (this was the predominant reason opening up their fabs last time failed).

By splitting manufacturing from design the IFS group effectively operates standalone and can treat internal and external similarly. It's possible, though unlikely, Intel could spin off foundry services after the new structure has all taken hold, but it's certainly not necessary for the business to grow.

 

[Kamen Rider Blade]

This post shows an impressive amount of ignorance as to how chip making/business works. No company looking at an external fab is going to be concerned with anything you're suggesting. Even if Intel were to steal something it would take them years to implement it, they would destroy their reputation and have nasty legal ramifications. The biggest actual concern is lack of access to leading edge nodes and Intel prioritizing its own products (this was the predominant reason opening up their fabs last time failed).

If you really believe the the world's chip clients are willing to trust IFS in it's current state, then you're more naive then I thought.
Look at how fast IFS is getting external customers.
Nobody has faith in IFS until they do a proper seperation.

By splitting manufacturing from design the IFS group effectively operates standalone and can treat internal and external similarly. It's possible, though unlikely, Intel could spin off foundry services after the new structure has all taken hold, but it's certainly not necessary for the business to grow.

It's going to be necessary if you want IFS to grow as a External Foundary, until then, I expect it to be slow and piddling as it is now.

 

[jamok99]

Wrong. Many credible reports including toms own article compared the node characteristics and clearly stated that Intel 4 is actually on par with TSMC N3 in many aspects (power, performance, density, etc). And the nextgen Intel 3 actually is not only on par with TSMC N3, but also exceeds it in certain aspects. Read the comparisons.

But this is all old news. Intel 3 is set be an "large" IFS node with a host of PDKs & libraries that'll cater not only to intel, but also to many other customers as well. Boring.

More recent exciting news is, Intel is coming out with it's high performance 20A node next year custom built for Arrow Lake only. And it's far far ahead of TSMC N3. Its a 2nm-class node with gate-all-around FET & backside-power-delivery. The first of it's kind in the industry. Beats TSMC 3nm by a mile.

(Note: TSMC N3 is based on old technology like Intel 3, Intel 4 & TSMC N5).

No offense, but do you remember the last time Intel actually delivered a crucial die shrink on time (that is, when they initially annouced the target date?) Funny, me neither.

 

[SiliconFly]

Intel's Fab won't be gaining much external customers until they seperate Intel (Chip Design) from their foundaries.

Similar to how TSMC has no (Chip Design) team, but just works as a "PURE Foundary" that contracts out to clients.

Until Intel is willing to split up the Foundary side from their (Chip Design) side and be seperate companies with no more ties, like AMD did with Global Foundaries, they'll always be bogged down by the (Chip Design) side and the lack of trust from working with their foundaries and if the foundary side will steal IP and knowledge to help the (Chip Design) side.

It's like we need a seperation of Church & State, but for (Chip Foundary & Chip Design).

You can do one only, you can't be both.

TSMC's entire business model is predecated on that seperation, Intel could stand too learn from TSMC and split themselves into two companies.
It'd be better for everybody around the world if they split up.

That's how you maintain trust.

Sadly, in a way, it's true. Apple moved away from Samsung to TSMC due to severe trust issues. Tim cook even praised that the security TSMC offers for their IP is unprecedented! Intel isn't a pure-play foundry like TSMC. Intel somehow needs to find a way to follow suit if their IFS needs to survive and thrive!

 

[SiliconFly]

But that's already happening. IFS has contracts with Amazon, Qualcomm, MediaTek, and the DoD - that we know of. Another announcement was for a major cloud vendor - likely either Amazon, Google, or Microsoft. Most of this appears to be on the future Intel 3 mode, while the DoD contract is for the 20A and later 18A nodes.

It will be interesting to see what starts coming out of these new fabs. This should start to show up sometime in 2024.

And lets be real, outside of Apple, nobody has access to TSMC N3.

So the foundry fight starts in 2024.

Apple Orders Entire Supply of TSMC's 3nm Chips for iPhone 15 Pro and M3 Macs

Apple has reportedly secured all available orders for N3, TSMC's first-generation 3-nanometer process that is likely to be used in the upcoming...

www.macrumors.com

I don't think the DoD contract is for 20A. Intel 20A is a lean and mean purpose-built node for it's own client group and not for external clients. But, Intel 18A is a full-fledged "fat" node (like TSMC N3 & Intel 3) with all the support tools, pdks & libraries designed for a wide range of customers.

 

[SiliconFly]

No offense, but do you remember the last time Intel actually delivered a crucial die shrink on time (that is, when they initially annouced the target date?) Funny, me neither.d

Rubbish. Intel has been in this business for more than 4 decades and has always been the market leader. They messed up only in 2017 with the 14++ nonsense. They've messed up so bad for a total of 5 - 6 years under the great leadership of bean counters & paper pushers who didn't have any foresight cos they didn't understand didly squat about technology! Idiots. And naturally, everything got messed up.

But that's history now. All the old nodes are gone for good! Intel 4 is ramping up very nicely as we speak and it can give a solid kick to TSMC N3 in power, performance & density! And the first batch of 14th-gen products are coming out in 6 months from now.

And the far more important news is, Intel 20A is almost ready with their test chips already out. The first arrow lake tape-in will happen anytime now! To put it in perspective, Intel is already in the advanced stages of the 20A implementation cycle with a tape-out in second half of this year and their 15th-gen arrow lake products out in late next year! Whereas, TSMC's competing node N2 is still in the drawing board and is expected to be available to its premium-paying tier-1 customers like Apple only in 2026!!! Tier-2 companies like AMD will get access to TSMC N2 only in 2027 (due to pricing).

In short, Intel is releasing its 2nm products in 2024 & AMD will have it's first 2nm products only in 2027! A gap of 3 years!!!

And by 2025, Intel will again move to 1.8nm leaving it's competitors in the dust. 5 nodes in 5 years. That's exactly what Intel CEO Pat Gelsinger meant last year when he said unquestioned leadership in 2025 & the competitors will be in the rear view mirror for good!

Trust me, AMD & TSMC have been the underdogs for decades. And they shined only when Intel tripped and fell flat in the face in the last 5 years. Now that Intel is back on track and running at a deadly pace, AMD's future is in question. I think AMD is headed for a disaster. If they don't do something different sooner rather than later, there's no future for AMD.