INTC - Intel

[Castanza]

 

Although quite lengthy, a pretty decent primer on the general industry.

 

1:00:00 Discussion on future of fab and foundry for Intel.

 

 

[Ballinvarosig Investors]

I am a complete imbecile when it comes to technology, but someone sent me this link today that explains the while ARM/x86 debate that is very pertinent to Intel.

 

 

Apple have already cut the cord to x86. My guess is that Google will follow next (they have no loyalty to any incumbent). Microsoft will not want to get left behind either of those two, it will be top priority for them to have an OS/services that run on ARM. Once those three are all focused on ARM, none of the smaller players will give a fcuk about x86. These three will have enough of a gravitational force to bring the industry with them.

 

I do not see a single player in the industry making a stand behind x86 and Intel. If anything, I think the opposite might be true, Intel were never a comfortable bed-fellow for anyone in technology from what I have read. With the x86/ARM battle close to over, the convergence of mobile and desktop is going to be closer to the mobile end (ARM) as opposed to the desktop side(x86).

 

I feel like Intel today is like Nokia back in 2010. Nokia posted record earnings after Apple came out with the Iphone. Even after the share price of Nokia was decapitated (in advance of terrible earnings), it was still worth a multiple of Apple. Look at Nokia and Apple in 2010, it's incredible to think Nokia was valued at a multiple to Apple. I feel we might be in a similar scenario today. End users want to be where the services are. They don't care about the devices themselves, let alone the internals. x86 is 60% of Intel's profit. I could be totally wrong, but I feel we are close to a top for record earnings for x86.

 

Before anyone says I have drunk the kool-aid, I say all this as someone who has owned various x86 laptops for the last 25 years and do not own an Android/Apple phone. I only need to look around me to realise that I am the outlier. Not a single person in my immediate cohort is buying x86 devices, my children are completely unable to operate a Windows computer. My mind is still blown after watching a three year old manage to unlock my wife's Iphone, opening youtube, and then search the first few letters to bring up a particular nursery rhyme video she wanted to watch.

 

x86 is the dinosaur that survived the meteorite strike, as far as I can see.

[lnofeisone]

NVDA looking to buy ARM from Softbank. Definitely an interesting move.

[Castanza]

NVDA looking to buy ARM from Softbank. Definitely an interesting move.

 

Very interesting. If this goes through, there could be big changes in partnerships. This industry is (imo) going to have some major shake ups over the next decade. Perhaps we will see new companies (Chinese) pop up. Two global manufacturers will not cut it long term. The supply constraint is already quite visible.

 

Currently only two major chip manufacturers (Intel, TSMC). Not counting Samsung.

 

NVDA was having some issues with TSMC negotiations. Highlighting both the positive and minuses of in-house foundry.

 

Apple uses ARM license to manufacture own products. I believe they are also working in their own discrete graphics card for laptops. If NVDA takes over ARM this could cause issues for Apple.

 

AMD is going to supply Apple in the near term with discrete graphics cards (in some models : likely to use Intel in higher end models). Apple said a few weeks ago they will move away from AMD in a few years.

 

Intel with its corporate issues managed to put out the Tiger Lake chip. Also utilizing more outsourcing for chips.

 

 

 

 

 

 

[Spekulatius]

NVDA looking to buy ARM from Softbank. Definitely an interesting move.

 

Very interesting. If this goes through, there could be big changes in partnerships. This industry is (imo) going to have some major shake ups over the next decade. Perhaps we will see new companies (Chinese) pop up. Two global manufacturers will not cut it long term. The supply constraint is already quite visible.

 

Currently only two major chip manufacturers (Intel, TSMC). Not counting Samsung.

 

NVDA was having some issues with TSMC negotiations. Highlighting both the positive and minuses of in-house foundry.

 

Apple uses ARM license to manufacture own products. I believe they are also working in their own discrete graphics card for laptops. If NVDA takes over ARM this could cause issues for Apple.

 

AMD is going to supply Apple in the near term with discrete graphics cards (in some models : likely to use Intel in higher end models). Apple said a few weeks ago they will move away from AMD in a few years.

 

Intel with its corporate issues managed to put out the Tiger Lake chip. Also utilizing more outsourcing for chips.

 

With all these mergers, the semiconductor industry which used to be extremely cyclical and competitive is becoming more oligopolistic. looks at consolidation in analog semis (Linear Tech, Maxim, ADI, memory , graphics ). Same with semi equipment suppliers where in some business lines, there is only one dominant supplier left ( lithographic - ASML,  inspection equipment - KLAC).

 

It has become a somewhat cozy business and the growing margins clearly show this.

[rkbabang]

 

Currently only two major chip manufacturers (Intel, TSMC). Not counting Samsung.

 

Global Foundries still exists (my group switched from TSMC to Global about two years ago, although my company as a whole uses TSMC more) and there are a lot of smaller internal company fabs around for older or specialty processes, which still produce a lot of chips and revenue. The company I work for still operates a number of such fabs.  For the cutting edge stuff you are correct, but the vast majority of semiconductor products do not need 10nm, 7nm, or 5nm ... processes.

 

[Liberty]

 

Currently only two major chip manufacturers (Intel, TSMC). Not counting Samsung.

 

Global Foundries still exists (my group switched from TSMC to Global about two years ago, although my company as a whole uses TSMC more) and there are a lot of smaller internal company fabs around for older or specialty processes, which still produce a lot of chips and revenue. The company I work for still operates a number of such fabs.  For the cutting edge stuff you are correct, but the vast majority of semiconductor products do not need 10nm, 7nm, or 5nm ... processes.

 

Yeah, people forget that there's a crapload of chips out there that don't need to be on the latest node... TI and ADI make a zillion of these things

[Castanza]

Sure that's fair

 

I guess I'm coming more from the cutting edge perspective. I mean this will be the driving force for companies like NVIDIA, Intel, AMD, Samsung etc.

[Liberty]

Btw, this won't cause any problems for Apple. Apple is a founding partner of the company and has a perpetual license. They use their own in-house design and don't use the Cortex designs made by ARM.

 

Jense also said that he plans to keep the open license model for ARM (he'd be crazy not to.. and would get antitrust trouble).

[Castanza]

Btw, this won't cause any problems for Apple. Apple is a founding partner of the company and has a perpetual license. They use their own in-house design and don't use the Cortex designs made by ARM.

 

Jense also said that he plans to keep the open license model for ARM (he'd be crazy not to.. and would get antitrust trouble).

 

Interesting

 

I was wondering about the potential anti-trust with this acquisition.

[rkbabang]

 

Currently only two major chip manufacturers (Intel, TSMC). Not counting Samsung.

 

Global Foundries still exists (my group switched from TSMC to Global about two years ago, although my company as a whole uses TSMC more) and there are a lot of smaller internal company fabs around for older or specialty processes, which still produce a lot of chips and revenue. The company I work for still operates a number of such fabs.  For the cutting edge stuff you are correct, but the vast majority of semiconductor products do not need 10nm, 7nm, or 5nm ... processes.

 

Yeah, people forget that there's a crapload of chips out there that don't need to be on the latest node... TI and ADI make a zillion of these things

 

Exactly.  High asking prices + cheap to manufacture = high margins.

 

[mwtorock]

Intel Corp. INTC +2.14% is nearing a deal to sell a memory-chip unit to South Korea’s SK Hynix Inc. 000660 1.64% for roughly $10 billion, in a move that would reorient the semiconductor giant away from an area of historical importance that has become increasingly challenged.

 

https://www.wsj.com/articles/intel-nears-deal-to-sell-nand-memory-unit-to-sk-hynix-11603129781?mod=hp_lista_pos2

 

Consolidation in NAND space. Maybe it is not that attractive to intel or they need the money for something else...

[Nelg]

In 2003-2006, AMD beat Intel to an x86-compatible 64-bit CPU and soon after that beat them to dual-core processors. As a result, from 2005-2006, Intel's desktop+server CPU segment's operating income fell by almost 2/3, and this was despite Intel having a manufacturing advantage vs AMD (which I don't believe had ramped up their 300mm fab yet).

 

Is anyone else worried something like this (or worse) could happen? What are the mitigants to this?

[Liberty]

https://mule.substack.com/p/intels-troubles-and-lam-researchs

[Spekulatius]

Anyone noticed that they actually met earnings and beat on revenues? The sale of their NAND business which lost hundred of millions for $9B is good news too - every little bit helps.

 

Back in 2006, when they were under pressure from AMD as well, they traded for >20x earnings. Apparently, expectations are much worse now.

[Jurgis]

I noticed that they are still on track to make 20B annual FCF as they were before. ./shrug

 

I continue to hold a position.

[OnTheShouldersOfGiants]

Is it just me or does X86 in general seem to be under attack?

 

Microsoft is further refining windows on ARM with 64bit compatibility, Apple moving Macs to ARM processors, AWS designing their own ARM silicone for the datacenter which they claim have 40% better performance for 20% less cost.

 

For those with expertise in the area, is it possible in your opinion that ARM can eventually take over most workloads currently served by X86 processors?

[beerbaron]

ARM will not take over server any time soon. Consider how sticky a software is to an entreprise, once implemented accross the group it's nearly impossible to get rid of it. Now, imagine all your infrastructure running each of those software on X86 and someone coming to say: "Wanna save on your electricy bill? All you have to do is convince all your software providers to compile into ARM, move all your data and servers...". Operationally there is no point to do all that (if possible) for saving a cost that probably is less than 1% or your operating expense.

 

What will happen is a word of x86 and a word of ARM living in parrallel.

 

Does anybody know how many chips are now released that use 10nm? Intel was saying they targetted 9 by the end of 2020 and that yields are above expectations. We are now near the end of the year, if they are close to delivering their 9 maybe that means the 10nm is now in order and ready to start competing with TSMC 7nm.

 

Thanks

BeerBaron

[Castanza]

I noticed that they are still on track to make 20B annual FCF as they were before. ./shrug

 

I continue to hold a position.

 

Tech seems to ebb and flow. Intel is out of favor like AMD was out of favor from 2005-2014ish. Sometimes the simplest thesis is the best. I think their fab facilities will become more crucial with time. They do have a culture issue and a leadership issue. But that is always fixable. I also still hold a position and will likely add a bit here and if it continues to down on negative sentiment.

[Spekulatius]

I noticed that they are still on track to make 20B annual FCF as they were before. ./shrug

 

I continue to hold a position.

 

Tech seems to ebb and flow. Intel is out of favor like AMD was out of favor from 2005-2014ish. Sometimes the simplest thesis is the best. I think their fab facilities will become more crucial with time. They do have a culture issue and a leadership issue. But that is always fixable. I also still hold a position and will likely add a bit here and if it continues to down on negative sentiment.

 

I have a small-medium sized position and consider adding after I looked at the release and the CC transcript. There really isn’t much unexpected bad news except shrinking ASP for server chips and it isn’t clear that it is an Intel specific issue.

 

Other than that, Intel has released some 10nm node (equivalent to 7nm TSMC) chips, but yields are still an issue , which isn’t unexpected (it was telegraphed last quarter).

 

Despite the above ,they met earnings and beat slightly on revenues.

 

The deal to sell the loss making (-900M last year) NAND Business is a strong positive.

 

No new issues with 10nm or their 7nm developed seem to have popped up.

 

The Bull thesis is that  Intel is still a leader in chip tech and trades cheaper than it did for eons. It is also a strategic assets for the US and the US government will do everything it can to make sure it remains so:

https://www.valueinvestorsclub.com/idea/INTEL_CORP/6955176147

 

 

It is clear that Intel has issues, and needs some serious adjustments but I actually think that the current CEO does the right things. It’s sure would be better if a tech wizard were still the helm, but for now, I don’t think he is doing a bad job.

[perulv]

Is it just me or does X86 in general seem to be under attack?

 

Microsoft is further refining windows on ARM with 64bit compatibility, Apple moving Macs to ARM processors, AWS designing their own ARM silicone for the datacenter which they claim have 40% better performance for 20% less cost.

 

For those with expertise in the area, is it possible in your opinion that ARM can eventually take over most workloads currently served by X86 processors?

 

ARM will not take over server any time soon. Consider how sticky a software is to an entreprise, once implemented accross the group it's nearly impossible to get rid of it. Now, imagine all your infrastructure running each of those software on X86 and someone coming to say: "Wanna save on your electricy bill? All you have to do is convince all your software providers to compile into ARM, move all your data and servers...". Operationally there is no point to do all that (if possible) for saving a cost that probably is less than 1% or your operating expense.

 

What will happen is a word of x86 and a word of ARM living in parrallel.

Thanks

BeerBaron

 

I am no expert in hardware or chip-design, or operating system development for that matter. But I do develop and architect software that "runs in the cloud". Like most modern "cloud native" systems, our software runs on a high level abstraction (think "serverless", "Platform as a service", various container-architectures etc) above both the hardware and OS. In addition, we utilize "managed" services whenever possible, for stuff like databases. Like most real software-solutions with some complexity, our software uses on a high number of such services (https://azure.microsoft.com/en-us/services/, similar for Amazon and so on). This enables us to do what we are good at, and let Microsoft do all the rest.

 

This is of course not something new, and the whole idea of abstractions/specialization since the start of computing/mankind. But my point is:

 

• For all those managed services, we don't even know what hardware they run on. We use a well defined interface to utilize this black box, and it could run on a rack of raspberry pies, Intel or AMD processors, ARM or X86.
  
• For our own code: "it depends". We use the Microsoft .NET Core platform, running in Microsoft Azure. .NET core supports ARM, but I think it would differ from case to case how easily you could run it on ARM instead of X86.
  

 

I would think that much more of the hardware-resources are spent by these services that are common to your online bank, dating site and streaming service, than on the software that makes them different.

 

We (and every other modern paas/saas solution) stand on the shoulders of the giants that are Microsoft (both as cloud provider and "framework provider"), Amazon etc. Our vendor-lock in is much more to this layer, than the hardware-layer. And for these few, it might be worth the time and effort to change a service from running on Intel to something else.

 

Of course, far from all current software is "cloud native", and I do not disagree that X86 and ARM will live in parallell. But at least for the modern architectures in the major clouds, there might not be such a high degree of vendor-lock in.

 

(All this said, I bought INTC a few months myself, and plan on holding)

[changegonnacome]

I get the strategic US asset arguement and its compelling - in a bifurcated world having on-shore design and manufacturing of semis etc. is possibly a national security issue of the highest order............. the question is what meaningful action could the US government do to support intel as a national champion in the semi space.....supporting more materials / semi research at a US academic level through NSF etc. is one. Guaranteed purchases another........I’m not sure either get Intel back in competitive lock step with TSMC.

 

How else could the US government ‘help’ Intel?

[Spekulatius]

I get the strategic US asset arguement and its compelling - in a bifurcated world having on-shore design and manufacturing of semis etc. is possibly a national security issue of the highest order............. the question is what meaningful action could the US government do to support intel as a national champion in the semi space.....supporting more materials / semi research at a US academic level through NSF etc. is one. Guaranteed purchases another........I’m not sure either get Intel back in competitive lock step with TSMC.

 

How else could the US government ‘help’ Intel?

 

There are a lot of options: direct funding for dual use (military and commercial) process development work. Guaranteed contracts for semis used in military applications. Preferred vendor etc. subsidies for corporation with US equipment vendors etc.

 

I am sure they can figure out something. In the end however, it is correct the government can support Intel, but nothing guaranteedPs success, especially if they become complacent, which is a real risk when working with and for Uncle Sam.

[Ballinvarosig Investors]

It's the end of the line for Intel on Apple platforms from next week, the new Macbook with an ARM processor was unveiled today.

 

https://www.theverge.com/2020/11/10/21552828/apple-macbook-air-arm-m1-chip-laptop-specs-price-release-date

 

Very impressive looking, increased performance, increased battery life, instant wake, no fan. Starting at $999. Runs iOS apps natively, Microsoft and Adobe will also be supporting it.

[no_free_lunch]

Between this, arm, scrappy amd, and lack of gpu products I don't see Intel's future as bright.  Not sure I would go so far as to short it but I am thinking about puts.  Might be a good way to hedge the market and could pay out regardless of market direction.  Of course I am probably wrong on timing and thus could play out over a decade.