News Intel beats EPS but misses on revenue in 4Q17

[Hitman928]

Will update post with more. Earnings call soon.

+Intel has guided downward on FY19 revenue. Guidance for $71.5B vs $73.1B expected. Data center expected up mid single digit and client side expected down low single digit.

+Desktop unit volume dropped 7% Q/Q and 8% Y/Y while ASPs rose 5% and 13%, respectively. Notebook units dropped 10% on the quarter and grew 1% on the year with ASPs up 3% and 6%. Data center units fell 3% Q/Q and grew 9% Y/Y with ASPs up 2% and 1%.

+Q4 revenue breakdown: Client Computing, $9.82B (consensus: $10.01B); Data Center, $6.1B (consensus: $6.35B); IoT, $816M (consensus: $812.7M); Non-Volatile Memory Solutions, $1.11B (consensus: $1.12B); Programmable Solutions, $612M (consensus: $518.7M); Other, $231M (consensus: $297.5M).

https://seekingalpha.com/news/3425818-intel-minus-7-percent-downside-fy-revenue-guide

On the plus side for investors, they increased their dividend by 5%.

 

[maddie]

I don't think size is the issue. Headcount has been roughly level for the last 10 years.

The processes at Intel are institutionally corrupt. Everything from how technical work is done to how people are treated. The things you do at Intel to further a career are considered liabilities elsewhere, this is not an exaggeration.

Even a new outsider CEO faces the insurmountable challenge of an entrenched lifer clique that is intent on milking the cow all the way to the end.

Intel will be nothing more than a distasteful backup to TSMC capacity before I retire. Bank on it.

A key statement. Entrenched culture resisting change.

The only effective driver is a true existential crisis. You see it in humans all the time. You only change when all other options are exhausted.

 

[dmens]

Not that I am privy to the details of Amazon's CPU but I highly doubt they would just take the ARM hard IP and use it as is.

When their very high infrastructure demands, Amazon is better off having their own custom CPU team (or even two or three teams) instead of paying an external vendor.

 

[Ajay]

If Intel goes down, I expect a fight between AMD, Marvell, Huawei, and possibly OpenPOWER cloners. And uh maybe Samsung?

If Intel 'goes down', I’d expect that x86 would be on the way out as well. So, AMD wouldn’t have a very bright future either.

 

[Markfw]

If Intel 'goes down', I’d expect that x86 would be on the way out as well. So, AMD wouldn’t have a very bright future either.

The problem is the installed software base. Its so large, I don;t think it will die in our lifetime. Every large business I have worked for (>100,000 employees) has possible over a million PC's each with some custom software, but all with huge amounts of commercial software.

 

[dmens]

x86 and Intel are not going to die and disappear, but if the revenue stream from x86 servers takes a severe hit then Intel cannot sustain itself in its current form. They will be forced to lease out their fab capacity, which is a tall order since their last attempt failed miserably.

The trend towards homegrown cloud infrastructure is a threat to AMD as well, but their exposure is much less severe.

 

[Ajay]

The problem is the installed software base. Its so large, I don;t think it will die in our lifetime. Every large business I have worked for (>100,000 employees) has possible over a million PC's each with some custom software, but all with huge amounts of commercial software.

With the push towards the web, cloud services and mobile, I’m beginning to think the x86 'golden handcuffs' aren’t so golden anymore. There are, for example, large shifts going on in software development languages and target platforms. The installed base can survive for some time while new sales increasingly focus on new paradigms which look to be increasingly service oriented.

 

[DrMrLordX]

When their very high infrastructure demands, Amazon is better off having their own custom CPU team (or even two or three teams) instead of paying an external vendor.

Why? The change over from OEM to ODM machines made sense, but I don't see how an internal CPU design team saves you money. Unless it really is just a supply issue. Everyone is going to be restricted by wafer supply, though. If Amazon (et al) are looking to save real money, the real killer is software licensing fees, though I expect they've made significant headway there already.

With the push towards the web, cloud services and mobile, I’m beginning to think the x86 'golden handcuffs' aren’t so golden anymore. There are, for example, large shifts going on in software development languages and target platforms. The installed base can survive for some time while new sales increasingly focus on new paradigms which look to be increasingly service oriented.

With what dmens is discussing, it looks like the major players are ready to produce their own software in-house, so ISA may not be so important. Lots of money to be saved there.

If Intel 'goes down', I’d expect that x86 would be on the way out as well. So, AMD wouldn’t have a very bright future either.

ISA will only matter so much. Also, AMD did originally design Zen as a companion to K12. There's little stopping them from taking a lot of what makes Zen/Zen+/Zen2 distinctive and applying those technologies to an ARM design.

 

[dmens]

Why? The change over from OEM to ODM machines made sense, but I don't see how an internal CPU design team saves you money. Unless it really is just a supply issue. Everyone is going to be restricted by wafer supply, though. If Amazon (et al) are looking to save real money, the real killer is software licensing fees, though I expect they've made significant headway there already.

Vendors charge a huge overhead per chip sold. Amazon deploys many million cores, it is a no-brainer for them to build their own, and on top of that, they will control their own destiny.

 

[DrMrLordX]

Vendors charge a huge overhead per chip sold. Amazon deploys many million cores, it is a no-brainer for them to build their own, and on top of that, they will control their own destiny.

By vendors do you mean OEMs, or chip manufacturers? Intel's margins are very high, but I would expect things to be different for up-and-comers that are trying to knock Intel down a peg.

 

[dmens]

By vendors do you mean OEMs, or chip manufacturers? Intel's margins are very high, but I would expect things to be different for up-and-comers that are trying to knock Intel down a peg.

Let's just say chip designers. Once the exorbitant Intel tax is out of the picture, hiring your own team to do what a fabless design house is doing makes more sense if you have the volume to justify it. Amazon certainly does.

 

[frozentundra123456]

Well, "Low post count Intel hater" is technically correct. But I have the credentials to emit that hate because of the many, many years spent inside the belly of the beast working on relevant products.

OK, disgruntled ex-employee then as well.

 

[frozentundra123456]

Nah man, he worked on Phi. He and Ctho95 - who was with AMD at the time - would have some very good technological discussions.

That was a different time...it was a time where you wouldn’t get a downvote from a moderator for asking where someone who hasn’t posted in a long time has been.

Well, the way the forums are these days, I consider a down vote a compliment.

 

[dmens]

OK, disgruntled ex-employee then as well.

Guilty as charged and very happy about the ex status. Feel free to actually argue against my take on industry trends instead of tossing ad hominems.

As for the honor of Intel, I've seen things that would shock your eyelids.

 

[moinmoin]

Let's just say chip designers. Once the exorbitant Intel tax is out of the picture, hiring your own team to do what a fabless design house is doing makes more sense if you have the volume to justify it. Amazon certainly does.

To add to this: Big cloud service providers often are essentially system designers anyway (designing all the server architecture from room scale to the single boards and all related details like cooling, interconnects etc.). From there the jump to including the CPU itself in the design planning isn't as far fetched as it first may seem.

 

[DrMrLordX]

Let's just say chip designers. Once the exorbitant Intel tax is out of the picture, hiring your own team to do what a fabless design house is doing makes more sense if you have the volume to justify it. Amazon certainly does.

Okay, I can accept that as a possibility. It's still up to Amazon's team to do a good-enough job to meet the needs of the organization. Paying a design team to produce a non-performant custom product would not strike me as a terribly good idea. If they can make something A76-based with strong interconnects then they'll be on the right track.

To add to this: Big cloud service providers often are essentially system designers anyway (designing all the server architecture from room scale to the single boards and all related details like cooling, interconnects etc.). From there the jump to including the CPU itself in the design planning isn't as far fetched as it first may seem.

That's basically what Huawei did (with help from their HiSilicon division which had experience developing mobile ARM CPUs). I still get the sneaking suspicion that Amazon may throw in the towel, though. History is filled with the wreckage of failed CPU design efforts. I would say that "time will tell", but in Amazon's case, how would we ever know? They probably wouldn't release specs on anything they designed strictly for internal use, and ODM sales are almost entirely "under the table". We won't know if they continue buying hardware from outside vendors, either.

 

[moinmoin]

I still get the sneaking suspicion that Amazon may throw in the towel, though. History is filled with the wreckage of failed CPU design efforts. I would say that "time will tell", but in Amazon's case, how would we ever know? They probably wouldn't release specs on anything they designed strictly for internal use, and ODM sales are almost entirely "under the table". We won't know if they continue buying hardware from outside vendors, either.

The ARM servers are A1 instances so we can watch how long they are available and how the prices compare to other instance types (assuming that they'd increase the price if they are deprecating them).

 

[CHADBOGA]

Guilty as charged and very happy about the ex status. Feel free to actually argue against my take on industry trends instead of tossing ad hominems.

As for the honor of Intel, I've seen things that would shock your eyelids.

Would love to hear about any details you are willing to share.

 

[realibrad]

Okay, I can accept that as a possibility. It's still up to Amazon's team to do a good-enough job to meet the needs of the organization. Paying a design team to produce a non-performant custom product would not strike me as a terribly good idea. If they can make something A76-based with strong interconnects then they'll be on the right track.



That's basically what Huawei did (with help from their HiSilicon division which had experience developing mobile ARM CPUs). I still get the sneaking suspicion that Amazon may throw in the towel, though. History is filled with the wreckage of failed CPU design efforts. I would say that "time will tell", but in Amazon's case, how would we ever know? They probably wouldn't release specs on anything they designed strictly for internal use, and ODM sales are almost entirely "under the table". We won't know if they continue buying hardware from outside vendors, either.

Hell, Amazon could just buy AMD and cut out the overhead.

 

[Arachnotronic]

Hell, Amazon could just buy AMD and cut out the overhead.

Do you think Amazon would be interested in running consumer CPU and GPU businesses as well as a game console/semi-custom SoC business? Buying AMD (or any of the Arm server companies that isn't a pure-play server SoC maker) would be problematic for the company.

Amazon made its bet by acquiring Annapurna Labs and I'm sure it's funding that asset handsomely to fuel its future server SoC ambitions.

 

[realibrad]

Do you think Amazon would be interested in running consumer CPU and GPU businesses as well as a game console/semi-custom SoC business? Buying AMD (or any of the Arm server companies that isn't a pure-play server SoC maker) would be problematic for the company.

Amazon made its bet by acquiring Annapurna Labs and I'm sure it's funding that asset handsomely to fuel its future server SoC ambitions.

It was more of a joke. If Amazon wanted to cut out the tax of profits, they already could quite easily.

Though, I must say that Amazon owning AMD would not mean that they would have to Run them in any real way. But, there are multiple problems as to why it would never happen.

 

[dmens]

Would love to hear about any details you are willing to share.

Heh, I think I'll leave it with the comments on institutional rot. Getting sued is not my idea of fun.

Let's just say that the most blistering criticisms on Blind and Glassdoor are accurate and on point.

 

[PingSpike]

Does AMD have some kind of poison pill clause with the x86 license in the event of a sale? It actually seems like some one should have bought them by now during their toilet years just to get at their patents.

I suppose you could just use a magical reverse buyout accounting trick to get around that though.

 

[DrMrLordX]

The ARM servers are A1 instances so we can watch how long they are available and how the prices compare to other instance types (assuming that they'd increase the price if they are deprecating them).

Hmm. That would require some creative interpretation of data, but it's better than nothing.

Hell, Amazon could just buy AMD and cut out the overhead.

Heh. Unlikely, but . . .

Does AMD have some kind of poison pill clause with the x86 license in the event of a sale? It actually seems like some one should have bought them by now during their toilet years just to get at their patents.

I suppose you could just use a magical reverse buyout accounting trick to get around that though.

We'll know what's really going on when Lisa Su announces that they've acquired Amazon in a tax inversion. Heaven only knows why Amazon would want an x86 license, though. It ain't worth what it used to be.

 

[Ajay]

We'll know what's really going on when Lisa Su announces that they've acquired Amazon in a tax inversion. Heaven only knows why Amazon would want an x86 license, though. It ain't worth what it used to be.

IIRC, if AMD gets bought out by another company, its cross licensing deal with Intel is voided.

 

[DrMrLordX]

IIRC, if AMD gets bought out by another company, its cross licensing deal with Intel is voided.

Yup, hence the tax inversion reference. The license may be sophisticated enough to trigger in the event of such shenanigans but who knows for sure?