Question Intel Q2 Results - Terrible

[jpiniero]

https://d1io3yog0oux5.cloudfront.net/_0f37de80c52d83ac6338b4805dea84aa/intel/db/887/8856/earnings_release/Q2+22_EarningsRelease+%281%29.pdf

Revenue down 22%.

Edit: Oh and it gets better... Intel posted a loss!

 

[TheELF]

Shouldn't that be one of the very first requirements of Intel? And it's confirmed it isn't?

Why though?! The chips act is about getting tech made in the usa, how is getting more tech from more companies made in the usa a bad thing? As long as a good part of it is also intel CPUs.
Also pat said that their investment would increase from 20 to 100 bil if they get some of that chip money so even if they spend all of the chip money only on actual CPU making there is still plenty more for IDM.

LOL might be???

This quarter was 15.3 ,looks pretty average for 2017.

Intel Revenue 2010-2023 | INTC

Intel annual/quarterly revenue history and growth rate from 2010 to 2023. Revenue can be defined as the amount of money a company receives from its customers in exchange for the sales of goods or services. Revenue is the top line item on an income statement from which all costs and expenses are...

www.macrotrends.net

2017-12-31	$17,053
2017-09-30	$16,149
2017-06-30	$14,763
2017-03-31	$14,796

It's already biting him, considering how furious Intel is at the Raptor Lake leaks. People have been turned off by how little improvement it brings compared to ADL. Not good when you are a few months away from launch. The leaks are just going to convince people to jump on AM5.

AM5 has zero increase in core count and only a small increase in IPC (as does raptor) the only thing heavily increasing on AM5 is the power consumption (and with that clocks) which until now everybody was praising as the best thing ever...
Raptor at least has a substantial increase in multithreaded workloads.
Depending on how the reviewers are going to spin this for maximum clicks, AM5 could be perceived as terrible perf/wat compared to raptor by the public.

I would prefer to see the space taken up by E-cores to be replaced with cache so they can have their own 5800X3D clone. But no, it would be too much for Intel to do the sensible thing.

Well you can only pull this trick one time and every other time after that it is going to seem much less impressive, so intel is saving this for when it will make the most impact, also they already tried it with broadwell so they might have insights and not do it at all.
Also also foveros might allow them to put cache above e-cores without loosing anything ,just like tsmc stacked it instead of replacing something else.

 

[DrMrLordX]

This quarter was 15.3 ,looks pretty average for 2017.

It's quite a different look when that revenue figure is part of a downward trend. Intel has no current enterprise CPU product until 2023. None. Their best server chip is vageuely competitive with Rome. That is not good for near-future DCG revenue. It's caught up with them, and it's eating DCG alive.

Client group has a current product, for now. Sadly CCG is losing even more money, more-likely due to downtrends in the economy and consumer demand issues.

 

[Markfw]

It's quite a different look when that revenue figure is part of a downward trend. Intel has no current enterprise CPU product until 2023. None. Their best server chip is vageuely competitive with Rome. That is not good for near-future DCG revenue. It's caught up with them, and it's eating DCG alive.

Client group has a current product, for now. Sadly CCG is losing even more money, more-likely due to downtrends in the economy and consumer demand issues.

When the AMD report comes out next week, we will see. If they are selling everything they can make (like before) Then I say in server its not s downward trend.

 

[Tuna-Fish]

This quarter was 15.3 ,looks pretty average for 2017.

You forgot to correct for inflation. It's been so flat for so long that it's easy to forget you even need to do it, but 2017q2 revenue in 2022 money is $17.8B. Today's results are down 15% from 5 years ago.

 

[jpiniero]

It also costs a lot more. 10 nm being what it is, plus the products are much bigger.

The original Coffee Lake-S was 149.6 for the bigger die and 126 for the smaller die. Alder Lake is 215 and 162.

And Raptor Lake is surely a lot bigger. You can see why the rumors that they aren't going to do a small Raptor Lake die and the bigger is probally what triggered that 20% price hike that's allegedly coming.

 

[coercitiv]

And maybe now that I spend a little bit less time in Washington, the focus for us as a team is getting that execution to be superb once again.
Pat Gelsinger , Q2 2022 Earnings Call

Goog luck with the new contra-revenue scheme!

 

[KompuKare]

And maybe now that I spend a little bit less time in Washington, the focus for us as a team is getting that execution to be superb once again.
Pat Gelsinger , Q2 2022 Earnings Call

Goog luck with the new contra-revenue scheme!

So he's managed to secure a largesse rarely seen outside of the military industrial complex, and he's complaining that it was a distraction?
It's almost like Intel and their shareholders expect money to grow on trees!

 

[coercitiv]

and he's complaining that it was a distraction?

I think his aim was to present the Washington win as a sign for future success in execution, but I also saw it as a slip of the tongue of sorts.

Based on this other answers in the Q&A, I'm curious to see what happens with Meteor Lake next year.

 

[Markfw]

So he's managed to secure a largesse rarely seen outside of the military industrial complex, and he's complaining that it was a distraction?
It's almost like Intel and their shareholders expect money to grow on trees!

What ? It HAS grown on trees for them for many years. That or they have been printing it.

 

[pakotlar]

care to expand?

Er, I misremembered. The **600K parts I thought went from 8 cores in 12600K to 16 in 13600K, but it’s 10 to 14. 13900K gets 24 vs 16 for 12900K. So 30-50% more cores (2x e-cores). Anyway, its a nice increase for a refresh part. I think U series laptops will get closet to 2x overall, from 2+8 max to 2+16, but I could be wrong.

 

[pakotlar]

You forgot the watts.

Yeah, though that increase on paper is very small; 50W lower PL4, 20W higher PL2, same PL1. In a similar boat is Zen 4, which actually increases both the PL1 and PL2 equivalents (TDP and PPT). Leaks definitely make it seem like Raptor Lake uses a lot more power in games, but I don’t trust those numbers, because the release notes for the bios version they tested on explicitly stated only basic compat, not performance optimized. Also, that leaker disabled the PL4 limit on Raptor and Alder Lake to get those peak power draw numbers.

 

[moinmoin]

Why though?! The chips act is about getting tech made in the usa, how is getting more tech from more companies made in the usa a bad thing?

Exactly, made not designed. The US already has plenty tech designed in the US, Intel's stuff, AMD's stuff, Arm high end cores, Apple Silicon etc. But Intel has the only leading edge fab in the US, CHIPS as I understand it should help the US to both stay in the leading edge fab race and allow more US designs to be made in the US as well. If Intel (is allowed to) put that money into its designs instead fabs it helps nobody but itself.

 

[TheELF]

Exactly, made not designed. The US already has plenty tech designed in the US, Intel's stuff, AMD's stuff, Arm high end cores, Apple Silicon etc. But Intel has the only leading edge fab in the US, CHIPS as I understand it should help the US to both stay in the leading edge fab race and allow more US designs to be made in the US as well. If Intel (is allowed to) put that money into its designs instead fabs it helps nobody but itself.

The IDM is external fabbing, intel fabbing designs of others, it's not a design farm for intel designs, it's a FAB for anybody to pay for to make their products.

Intel’s ‘IDM 2.0’ Strategy Defined in 60 Seconds

Intel has always been an integrated device manufacturer. In 2021, the company expanded its plans and capabilities.

www.intel.com

In March, CEO Pat Gelsinger introduced “IDM 2.0,” a major evolution of that strategy. Intel’s new IDM model includes significant manufacturing expansions, plans for Intel to become a major provider of foundry capacity in the U.S. and Europe to serve customers globally, and expansion of Intel’s use of external foundries for some of its products.

 

[DrMrLordX]

When the AMD report comes out next week, we will see. If they are selling everything they can make (like before) Then I say in server its not s downward trend.

I was referring to a downward trend for Intel specifically. They're losing more and more revenue every quarter. As for AMD, I expect their enterprise&embedded results (or however they organize it now) to be quite healthy in comparison. AMD may also suffer from losses in client but we will see. Different company, different topic.

 

[TheELF]

I was referring to a downward trend for Intel specifically. They're losing more and more revenue every quarter.

Every quarter here being this one and the previous one, two quarters total, while they had 10 quarters before that all between 18 and 19 bil...up from several years before that of 15-16 bil per quarter.
Also historically the 2nd and 3rd quarters are very often lower then the other two quarters, for both companies.

 

[Markfw]

Every quarter here being this one and the previous one, two quarters total, while they had 10 quarters before that all between 18 and 19 bil...up from several years before that of 15-16 bil per quarter.
Also historically the 2nd and 3rd quarters are very often lower then the other two quarters, for both companies.

You are missing the point. Q1 was not good, Q2 for them was terrible, and the real problem is that they have no products (server and hedt mostly) to compete with AMD, who for the last 2 quarters was bringing home the cash in huge numbers. If AMDs Q2 report that comes out Tuesday is better than their Q1, its a trend, and not a good one for Intel. With no competing parts until at least 2024, Intel will keep losing market share and money until that time.. I have no idea what 2024 will bring at this point.

 

[moinmoin]

The IDM is external fabbing, intel fabbing designs of others, it's not a design farm for intel designs, it's a FAB for anybody to pay for to make their products.

Intel’s ‘IDM 2.0’ Strategy Defined in 60 Seconds

Intel has always been an integrated device manufacturer. In 2021, the company expanded its plans and capabilities.

www.intel.com

You are thinking of IFS. IDM (like your link also states) is absolutely including internal designs (it stands for integrated device manufacturer after all) as well as outsourcing to eternal fabs like TSMC's. IDM shouldn't profit of CHIPS, whereas IFS should.

 

[DrMrLordX]

Every quarter here being this one and the previous one, two quarters total, while they had 10 quarters before that all between 18 and 19 bil...

Irrelevant, they're going back down for reasons which are troubling.

 

[Markfw]

Irrelevant, they're going back down for reasons which are troubling.

Exactly. It does not matter what they did for YEARS, it matters what they are doing now, and what they are going to do. And it does not look promising. TheELF just keeps making excuses for them, but it won't save Intel....

 

[IntelUser2000]

You are thinking of IFS. IDM (like your link also states) is absolutely including internal designs (it stands for integrated device manufacturer after all) as well as outsourcing to eternal fabs like TSMC's. IDM shouldn't profit of CHIPS, whereas IFS should.

IDM 2.0 is both. Pat said that the foundry model helps IDM stronger and vice versa. On one of his interviews he also said that being an IDM created issues that caused problems in the long run and he wants to get the company out of the mindset.

Nvidia has expressed interest in their fabs so if they are good enough they'll get customers.

If you read articles about how Intel operated you can see why no one cared about their previous foundry efforts. Before it was their to boost only themselves. But you have to cater to your customers. Some like Semianalysis talks about how Intel even treated even semi equipment vendors badly, since they thought themselves the king of the world.

It doesn't matter if your products are number 1, if your attitude and treatment of your customers are terrible. The hope they see with the new management is that they are doing lots to fix that.

Jeez, CEO salaries are obscene. I am just an armchair engineer but a part of me feels like after 10 million per year, CEO salary simply does not scale up with their skill and/or talent. I mean let's be real: Pat could have worked 24/7 for every week of his current tenure as CEO of Intel but I highly, highly doubt those extra hours could have averted the disaster which was the Q2 Earnings Report. There's simply so many engineers at Intel who are far more important to Intel's execution, or lack thereof, than Pat Gelsinger. If Pat and the executive team wrote down the 5 year game plan, which includes laying out particular milestones along the way, and then dipped out of there after a month, who's to say someone else couldn't guide the ship?

CEO pay is too much I agree. It's unfortunately part of the system which won't change until a titanic change of mindset occurs. Like when China becomes a superpower in a few years(after years of civil unrest around the world).

The previous CEOs fired many employees. Gelsinger's team hired 12,000 of them, mostly in engineering. Not only that the hiring and firing policy is that if you have 20 years of work in the company and you get promoted, after you do bad you'd get fired, which is an insane policy to me. You essentially tell the rest of the employees your loyalty is worth a crap. That's why Gelsinger lost in the first place, by Otellini. Why didn't he get demoted instead?

He's policy at previous places show a way different mentality. If anyone can change it around it's him. The question is whether the problem is a hard fix or an impossible one.

 

[moinmoin]

IDM 2.0 is both. Pat said that the foundry model helps IDM stronger and vice versa. On one of his interviews he also said that being an IDM created issues that caused problems in the long run and he wants to get the company out of the mindset.

Nvidia has expressed interest in their fabs so if they are good enough they'll get customers.

If you read articles about how Intel operated you can see why no one cared about their previous foundry efforts. Before it was their to boost only themselves. But you have to cater to your customers. Some like Semianalysis talks about how Intel even treated even semi equipment vendors badly, since they thought themselves the king of the world.

It doesn't matter if your products are number 1, if your attitude and treatment of your customers are terrible. The hope they see with the new management is that they are doing lots to fix that.

I'm well aware of all that. TheElf was responding claiming IDM is not about Intel designs which is obviously wrong, Intel designs of course are still part of IDM. It's IFS which is not about Intel designs but fabbing designs of others. That IDM is essentially Intel's new overarching business plan of which IFS is a part of is obvious, but that shouldn't change that CHIPS support goes only to the IFS part.

 

[pakotlar]

IDM 2.0 is both. Pat said that the foundry model helps IDM stronger and vice versa. On one of his interviews he also said that being an IDM created issues that caused problems in the long run and he wants to get the company out of the mindset.

Nvidia has expressed interest in their fabs so if they are good enough they'll get customers.

If you read articles about how Intel operated you can see why no one cared about their previous foundry efforts. Before it was their to boost only themselves. But you have to cater to your customers. Some like Semianalysis talks about how Intel even treated even semi equipment vendors badly, since they thought themselves the king of the world.

It doesn't matter if your products are number 1, if your attitude and treatment of your customers are terrible. The hope they see with the new management is that they are doing lots to fix that.



CEO pay is too much I agree. It's unfortunately part of the system which won't change until a titanic change of mindset occurs. Like when China becomes a superpower in a few years(after years of civil unrest around the world).

The previous CEOs fired many employees. Gelsinger's team hired 12,000 of them, mostly in engineering. Not only that the hiring and firing policy is that if you have 20 years of work in the company and you get promoted, after you do bad you'd get fired, which is an insane policy to me. You essentially tell the rest of the employees your loyalty is worth a crap. That's why Gelsinger lost in the first place, by Otellini. Why didn't he get demoted instead?

He's policy at previous places show a way different mentality. If anyone can change it around it's him. The question is whether the problem is a hard fix or an impossible one.

Could you elaborate on Gellsinger being fired? I thought he left when he was passed over for CEO?

 

[poke01]

now that big.LITTLE Intel-style is the next best thing ever, Intel have dropped AVX-512 altogether (from mainstream).

Intel's version of big.LITTLE is to win in Cinebench. Intel E-cores are for area efficiency, not power efficiency.

ARM's and Apple's E-cores are actually E-cores and Apple's E-cores are super efficient.

 

[Saylick]

Could you elaborate on Gellsinger being fired? I thought he left when he was passed over for CEO?

According to this article, "Gelsinger was pushed out in 2009 after he was blamed for the failure of Larrabee, an Intel effort to create a GPU that every analyst said was doomed to failure and not his fault. He did a three year stint as COO of EMC before taking over VMware in 2012."

 

[CHADBOGA]

Is the problem of making smaller manufacturing processes for semi-conductors so difficult, that maybe only a handful of people on the planet are capable of it and Intel doesn't currently have any of these people?

I have tried to ask this question before and maybe I didn't word it correctly, but the answers I got were that getting the manufacturing processes right is just a matter of the right management and being sufficiently resourced. Maybe that use to be the case, is it still?

Considering for how long Intel has had issues with their fab processes, from once being an untouchable leader to now being behind TSMC, what explains this?

Surely Intel have thrown insane amounts of money at their fab processes and revamped the teams working on the fab processes, but they seem stuck in limbo, barely able to progress in a timely manner.

So do TSMC have a couple of geniuses in the field of semi-conductor manufacturing that Intel lack and that is making all the difference?