Intel reports Q1 earnings: $12.6 billion revenue, $2 billion net income

[RampantAndroid]

Themarketinghype around "performance per watt" for a desktop workstation is just that... marketing hype.

The idea that doubling the power consumption of a workstation would cost a person "money" is patently absurd. We are talking pennies here. We are talking about the equivalent of a lightbulb or two when the cpu is going full throttle. It is laughable joke that Intel's marketing department is attempting to redefine performance as "performance per watt" for desktop users... this is beyond absurd and embarrassing.

Bake one batch of chocolate chip cookies in an electric oven, toast a slice of bread, dry a load of clothes in the dryer, blow dry your hair... now you've just trivialized the power consumption of a CPU. "performance per watt" for desktop use is just marketing desparation.

I work in a building with 300 people in it. You tell me that 100W per workstation is nothing. When the AC failed for the building, everyone ran for the hills.

You tell that to people setting up servers that already have hefty cooling requirements. A lab in an old building I was would rise 10 degrees (I assume F, but he might have meant C) for every 10 minutes the AC was off. Which happened a little too often because people would microwave things that set off the fire alarm, shutting down ALL air to the building and leaving it off until the fire department cleared the building. Servers started to shut down due to temps.

 

[colonelciller]

I work in a building with 300 people in it. You tell me that 100W per workstation is nothing. When the AC failed for the building, everyone ran for the hills.

yet another misunderstanding of the scale my friend. the CPU contribution to your building's heat is roughly equivalent to 300 light bulbs. bake one batch of chocolate chip cookies and you have trivialized the output of those CPUs. perhaps someone was baking a batch of toll house and that is what caused everyone to flee for the hills. or just maybe the SUN was shining???

You tell that to people setting up servers that already have hefty cooling requirements. A lab in an old building I was would rise 10 degrees (I assume F, but he might have meant C) for every 10 minutes the AC was off. Which happened a little too often because people would microwave things that set off the fire alarm, shutting down ALL air to the building and leaving it off until the fire department cleared the building. Servers started to shut down due to temps.

servers are special case scenarios aren't they. server room usage scenarios parallel open office workstation/home desktop usage scenarios how exactly?


translation...

 

[colonelciller]

The problem for you is that your wishes belong to a niche. And thats not what the huge wast majority of people wish.

So dont try and excuse it on something else. People simply voted with their wallet what they want. And they didnt want the same as you in the price range you are willing to pay. Its you being the strawman.

the only problem here is with distasteful marketing practices attempting to redefine performance to equal "performance per watt" which is utterly and completely absurd for home desktops and office workstations. the marketing scam being peddled by intel and people in this thread relies solely on people not understanding how much electricity.

my response to this thread about performance per watt (to those peddling it as meaningful) was precisely focused on how performance per watt is a ludicrous marketing tactic when used to peddle desktop and workstation CPUs. That was a focusced response to your performance per watt comments... that inot wfhat a straw man is.

the straw-man is your attempt to shift the focus of the discussion onto "niche desires" when the performanc per watt for desktops is shown to be without merit... the electricity cost savings that you touted as meaningful would never be noticed by any home user. the other straw man was the counter argument about whathappens inside a building with 300 people in it when the sun shines on it without electricity.

http://www.youtube.com/watch?v=v5vzCmURh7o

 

[ShintaiDK]

the only problem here is with distasteful marketing practices attempting to redefine performance to equal "performance per watt" which is utterly and completely absurd for home desktops and office workstations. the marketing scam being peddled by intel and people in this thread relies solely on people not understanding how much electricity.

my response to this thread about performance per watt (to those peddling it as meaningful) was precisely focused on how performance per watt is a ludicrous marketing tactic when used to peddle desktop and workstation CPUs. That was a focusced response to your performance per watt comments... that inot wfhat a straw man is.

the straw-man is your attempt to shift the focus of the discussion onto "niche desires" when the performanc per watt for desktops is shown to be without merit... the electricity cost savings that you touted as meaningful would never be noticed by any home user. the other straw man was the counter argument about whathappens inside a building with 300 people in it when the sun shines on it without electricity.

http://www.youtube.com/watch?v=v5vzCmURh7o

Offices want performance/watt for desktops. Due to TCO, noise and indoor climate. So dont tell me its some kind of PR stunt.

You simply seem agry because you are not getting what you wish on the desktop, within the price you are willing to pay. But yet you fail to understand how tiny a niche you really belong to.

You can keep talking about others as strawmen. But its you being the one.

Here is a 150W TDP Desktop CPU for you:
http://ark.intel.com/products/70845...ssor-Extreme-Edition-15M-Cache-up-to-4_00-GHz

 

[Cerb]

this is a straw-man and is missing the point completely.

intel is peddling performance per watt because intel is currently focused exclusively on mobile platforms where performance per watt is actually meaningful. problem is that for desktops this means that cpu power is flatlining, and the marketing nutjobs have decided to redefine performance as "performance per watt" simply because intel is no longer (capable of??) producing meaningful performance improvements for desktop users.

your comment about several kilowatts is a really nice attempt at straw-manning the discussion though... ...that is unless you're imagining CPUs as the integrated heating elements in a kitchen oven.

Either the several kW is OK, or it's not a strawman. We used to get increased power consumption along with increased performance. This was going on right into the P4 era. It stopped, and began reversing, before mobile was everything. Why?

Fact is, it's not pennies. It's anywhere from tens of dollars per computer to over a hundred dollars per computer (not in the U.S., typically, yet), per year. The difference isn't just your PC at your desk. It's your whole office, your whole server room, etc..

Computers are not unique, in this matter. Basically everything but electric ovens and stovetops have been actively getting more efficient, as both the awareness of resource waste has risen, and the effects of added machine noise and electricity costs have become more apparent and pressing. In the U.S., we're mostly still expecting $.20/kWh (some states are at that or higher, already, though), and it is starting to change habits. Businesses have cared a lot since the K8 offered higher performance without the higher bills.


Take the cookie baking. How often do you do that? It will take around 1kWh, with an electric oven, plus or minus some (1.5-2 if you wait the full pre-heat time, maybe). 1kW constant is 168kWh/wk. I'll go with 2kWh for pessimistic numbers.

                                cost/yr at kWh rate of
              | power draw | $.10 | $.15 | $.20 | $.25 | $.30
----------------------------------------------------------------
cookies/batch | 2kWh/batch | $.20 | $.30 | $.40 | $.50 | $.60
25W  average  |  219kWh/yr | $22  | $33  | $44  | $55  | $66
50W  average  |  438kWh/yr | $44  | $66  | $88  | $110 | $132
100W average  |  877kWh/yr | $88  | $131 | $175 | $219 | $263
----------------------------------------------------------------
Batches per year equivalency:
25W avg:  110
50W avg:  219
100W avg: 438

You can draw your own value conclusions, but it's not hard to see how aggregate average power savings can add up to real money, even without sky-high electricity costs.

 

[Haserath]

Isn't anybody getting bored of the same squabbles over and over and over... There is no true winner. You just find out which one is more determined to keep their stream up longer.

 

[Idontcare]

Bake one batch of chocolate chip cookies in an electric oven, toast a slice of bread, dry a load of clothes in the dryer, blow dry your hair... now you've just trivialized the power consumption of a CPU. "performance per watt" for desktop use is just marketing desparation.

This is all Intel's invention?

Did you totally miss the decade where AMD touted their superiority in desktop and server's for performance/watt over Intel when SOI was giving them an edge?

Intel doing the same with FinFet's a decade later suddenly makes it an Intel invention of "Themarketinghype"?

Bake one batch of chocolate chip cookies in an electric oven, toast a slice of bread, dry a load of clothes in the dryer, blow dry your hair... now you've just trivialized the power consumption of a CPU. "performance per watt" for desktop use is just marketing desparation.

1W of continuous power usage @ $0.115/kWHr nets out to costing you an extra $1/year on your power bill. (not including additional electricity for more A/C if applicable)

100W of more power costs me an extra $100 per year.

I'm not worried about the cost of baking a batch of chocolate cookies. I am worried about the opportunity cost of paying for that extra 100W over the course of a year because that $100 could take me and my entire family to Red Robin for a nice dinner (burgers and bottomless fries :thumbsup not once, but twice.

Now would I rather take my family to Red Robin twice over the course of a year, or would I rather have that money leak out of money checking account, $8 a month, to the power company?

Now multiply that 100W by the fact that I've got a family of four here and currently only two of us computer-users...but that will double to all four in a few years.

The difference between leaky power-sucking computers versus power-sipping computers in my house is going to start making a difference to the tune of hundreds of dollars in potential opportunity costs, I can send that money to the power company, or I can save it for a rainy day by simply making power-conscience decisions when buying my desktop hardware.

(and yes I apply that thinking to each and every electrical device I put into the house, from the lights to the fridge to the heat pump)

I suppose you probably find yourself loathe to this whole "get higher gas mileage" craze too. People should just drive V10-powered cars and get their 8mpg and think about baking cookies as their excuse for doing it. Why drive something that gets you better gas mileage? Its only money, who cares, right?

 

[Haserath]

Would Intel's prowess in fabs translate to any other industry?

I could imagine some kind of nano-imprinted battery from their fabs. If only it were a possibility...

They seem to be nearing the asymptote for profit on processors, because I don't think they'll ever break far into mobile.

 

[Phynaz]

Themarketinghype around "performance per watt" for a desktop workstation is just that... marketing hype.

You're thinking small. Now multiply that times 10,000 or 50,000 or 100,000 for a company. Not so much hype anymore, is it?

 

[djgandy]

You're thinking small. Now multiply that times 10,000 or 50,000 or 100,000 for a company. Not so much hype anymore, is it?

100,000 machines, that's not small either. In a power to initial investment ratio it is still that same. This is why we have percentages You could save more money by not replacing employee machines until a day later.

 

[krumme]

Why is it an obvious problem? Will Intel be involved in a multi-year covenant-heavy debt issuance worth many times their yearly EBITDA that will need cash flows from the new assets to be repaid? Because this is what I understand as being a CAPEX problem, and there is nothing of the sort going on with Intel. Intel doesn't need the cash flows of those new assets to pay for them, they need them to keep current ROI levels and fund future R&D levels.

And what are the choices anyway? Reduce CAPEX, slow down their road map and put a more conservative approach to the mobile market just to keep nicer financial metrics and stronger balance sheet but weaker overall market position? Because what you are suggesting will yield just that, without any of the benefits of their current strategy, trade off long term benefits for short term financial metrics.

And except in tail-risk scenarios, if their strategy backfires, in order to correct the "huge" CAPEX problem they will have to do more or less the same choices I pointed out in the above paragraph, except that they might have a weaker balance sheet. Instead of do this now, they are making a bet on their business model and on their company and betting everything on themselves, and if things go wrong, they will make those trade offs you and Pablo are expecting of them.

No its not "life and death" or "in a multi-year covenant-heavy debt issuance worth many times their yearly EBITDA"

We cant imho interprete capex out of context, meaning here change in the predicted benefit, and the ability to alter the capex with changing conditions.

We have been over this before, i am not saying Intel had a choice here. The risk at present absolutely looks manageable. But if the economy continues to be in a slump, notebook mainstream prices continues to fall, and their profit for 2013 looks to be less than predicted, they will slow down. We have to remember that they then also miss what they did not chose to do! - fx. expanding b2b market solutions leveraging on all the strong positions they have here fx. security.

The competition - that they have chosen together with the capex - is not what they have faced earlier in history. This look like a frontal attack on Samsung and Arm ecosystem. Thats good, if their main business is strong, and the competitors is comparable weak. Intels business is not in top shape now as i can see it. The server business will continue to generate stable amount of huge profit, but the consumer market is weak. The money goes to phones, screens whatever, just not pc. The money goes the other way than Intel Capex. Thats what the numbers tell me.

Going in the mobile consumer market does not loke like a short 3 years transistion to me either. The cpu/apu does not play the major part it does in especially server, but to lesser degree in consumer pc and notebooks. They can not control it the way they could.

 

[Blandge]

notebook mainstream prices continues to fall,

I thought notebook ASPs when up overall the last couple quarters?

 

[pablo87]

I think you are confusing capex with depreciation.

Regardless of Intel's depreciation "policy", the fact remains that their overwhelming investment is building fabs and their main business is selling processors - any which way you slice it, Intel doubled their investment in 2011, 2012 and projected for 2013, and the unit shipments have declined. Therefore, the real cash cost per processor of this investment has more than doubled.

Pretax Income of $2.5B equates to "cash income" of $2B adjusted for capex/depreciation in Q1. If we then apply a 27% tax rate, net income becomes $1.4B (also, Intel now pays instead of earns interest). Capex in Q1 was quite a bit lower than annual guidance/4 so if anything, real income will be even lower in upcoming quarters.

All in the face of declining volumes and stronger demand for lower ASP products.

 

[mrmt]

No its not "life and death" or "in a multi-year covenant-heavy debt issuance worth many times their yearly EBITDA"

We cant imho interprete capex out of context, meaning here change in the predicted benefit, and the ability to alter the capex with changing conditions.

We have been over this before, i am not saying Intel had a choice here. The risk at present absolutely looks manageable. But if the economy continues to be in a slump, notebook mainstream prices continues to fall, and their profit for 2013 looks to be less than predicted, they will slow down.

So I think we can agree that CAPEX isn't the fundamental issue here, as it is clear that they will execute their CAPEX plan, but a more a question of business model, and risk management, right?

Going in the mobile consumer market does not loke like a short 3 years transistion to me either. The cpu/apu does not play the major part it does in especially server, but to lesser degree in consumer pc and notebooks. They can not control it the way they could.

I'm not saying it will be completed in 3 years, I'm saying that in three years Intel should yield some results of the transition, if they don't, they are done for. Those things just take time, but we can have some signs on what the future will look like.

Take for example Intel. The current situation we have now for Intel was laid out fundamentally around 2004, when Conroe was in the design phase and Tick-Tock was being internally developed. In 2007 we could already see that Intel was on a right track.

 

[Idontcare]

Regardless of Intel's depreciation "policy", the fact remains that their overwhelming investment is building fabs and their main business is selling processors - any which way you slice it, Intel doubled their investment in 2011, 2012 and projected for 2013, and the unit shipments have declined. Therefore, the real cash cost per processor of this investment has more than doubled.

Pretax Income of $2.5B equates to "cash income" of $2B adjusted for capex/depreciation in Q1. If we then apply a 27% tax rate, net income becomes $1.4B (also, Intel now pays instead of earns interest). Capex in Q1 was quite a bit lower than annual guidance/4 so if anything, real income will be even lower in upcoming quarters.

All in the face of declining volumes and stronger demand for lower ASP products.

To whatever extent your argument is true, the Intel's gross margins do not reflect a "real cash cost per processor of this investment has more than doubled" outcome.

Somehow, and I am guessing this is where them being experts at what they do versus you just guessing at what they ought to be doing is making all the difference in a place we call "reality", Intel is managing their Capex outlays just fine sans the pablo87 consulting inputs.

 

[krumme]

I thought notebook ASPs when up overall the last couple quarters?

You are probably right. What i was referring to is what i understand was and is Intels own definition of a mainstream notebook price. It seems to creep down.

 

[Ajay]

No, it doesn't. Performance per Watt has been increasing in both CPUs and GPUs, and most GPUs have gotten improved performance at relatively stable TDPs, if not slightly decreasing over time.

If performance at the same power usage increases, you have improved performance per Watt. If performance improves by a greater amount than power is increased, you also have an improvement in performance per Watt. You can have your cake and eat it, too, to a fair degree.

If you offered one of those businesses a workstation with 20% more performance than they have now, but 4x the power consumption, do you really think they'd buy it? If performance efficiency were normalized to Core 2 era CPUs, and G92-based Quadros, that probably wouldn't be far off the mark.

If it didn't matter, we would indeed have individual workstations pulling several kW, by now. But, it does matter. You want more performance, but not only more performance. Power consumption staying the same, not improving at all, will mean you'll need increased revenues, no raises, and/or cutting of pay, over time, just to break even. Like everything else, electricity costs keep going up. They are not unimportant. Also, you only need a few space heaters before your engineers start caring about idle power consumption. Prescotts and Quadro FX cards got those well-paid engineers caring, even not themselves bearing the cost.

Either you didn't read what I wrote, or we have been exposed to very different engineering environments. ASIC designers cannot get enough compute power, they will use all of it and want more. They had the best workstations in the company as well as dedicated server farms for design and verification. When I was working on multi-million LOC projects, re-compiling individual directors was a piece of cake. Compiling and especially linking, the entire build was a waste of my time when I was working in critical time (near release and when a large customer had a serious bug) and had no other project that I was supposed to be working on during that time.

And while electricity wasn't unimportant, it paled in regards to market timing (bleeding edge high tech firm). I probably still have some ringing in my ears from the damn 15K HDs running in the 3 Sparc workstions that I had in the lab. I would stay late, just so that I could build my next test image on the beefier Sparcs in our "build room", which was just an oversized closet with beefed up AC, without having to compete with other engineers for resources. Management was more than happy to provide us with all the AC we needed in the lab and offices.

When was the electricity not unimportant - when it went out. And our managers were fretting because they were losing tens of thousands of dollars per hour to lost productivity.

Now the code bases are much larger, I haven't had the pleasure of working on projects with tens of millions of LOC (mostly Java desktop apps since and some Flex), but I am curious what those firmware engineers and ASIC designers are using today, when I left, the ASIC designers were getting even faster Sparcs and larger farms and firmware was moving over to PCs.

 

[Phynaz]

100,000 machines, that's not small either. In a power to initial investment ratio it is still that same. This is why we have percentages You could save more money by not replacing employee machines until a day later.

I manage the infrastructure for 33K machines. That one day delay will cost thousands of dollars in repairs for the roughly 45 machines a day that will have an age related hardware failure.

 

[Blandge]

You are probably right. What i was referring to is what i understand was and is Intels own definition of a mainstream notebook price. It seems to creep down.

I think that's the point of the Ultrabook, and it looks like it's working.

 

[Phynaz]

Regardless of Intel's depreciation "policy", the fact remains that their overwhelming investment is building fabs and their main business is selling processors - any which way you slice it, Intel doubled their investment in 2011, 2012 and projected for 2013, and the unit shipments have declined. Therefore, the real cash cost per processor of this investment has more than doubled.

Pretax Income of $2.5B equates to "cash income" of $2B adjusted for capex/depreciation in Q1. If we then apply a 27% tax rate, net income becomes $1.4B (also, Intel now pays instead of earns interest). Capex in Q1 was quite a bit lower than annual guidance/4 so if anything, real income will be even lower in upcoming quarters.

All in the face of declining volumes and stronger demand for lower ASP products.

One, you are mixing capex and expense.

Two, net income already has the taxes and depreciation removed , you are confusing net with EBITA.

Three, Intel lowered the capex expectations just because they will require less capex to ship less chips.

I'll put it in a much harsher phrase than IDC did - you pretty much have no clue what you are posting about.

 

[Ajay]

Three, Intel lowered the capex expectations just because they will require less capex to ship less chips.

In this quarterly report (guidance), I haven't read it yet?
In the last, or second to last quarter for which Intel provided guidance, CAPEX was expected to go up because of the aggressive conversion of older fabs to smaller nodes (specifically 14nm, IIRC).

 

[Phynaz]

Yes, Intel lowered their capex forecast for the year by $1B.

 

[mavere]

I thought notebook ASPs when up overall the last couple quarters?

The low-end moved from netbooks to tablets. I'm not sure if there was much up-selling to make that ASP increase a 'good' ASP increase, as opposed to lower revenues, even lower volume, so therefore higher ASP.

 

[VirtualLarry]

Would Intel's prowess in fabs translate to any other industry?

I could imagine some kind of nano-imprinted battery from their fabs. If only it were a possibility...

They seem to be nearing the asymptote for profit on processors, because I don't think they'll ever break far into mobile.

Intel should make solar cells.

 

[Blandge]

The low-end moved from netbooks to tablets. I'm not sure if there was much up-selling to make that ASP increase a 'good' ASP increase, as opposed to lower revenues, even lower volume, so therefore higher ASP.

I was under the impression netbooks weren't included in notebook sales.