that's not year over year,, thats just the quarter.. Its not as bad as you might think.
Intel's Performance cpu's are just now starting to use 5w of power... imagine in in a couple of years when 10nm is out and you have core i5 performance in a fanless 5w package. And the low power stuff can be used in phones at 2.5w
This doesn't exactly exude confidence from a fab perspective or even longer term performance. Not when Samsung and TSMC are increasing spending and even exceeding Intel.
Stacy Smith: "Actually, it’s a pretty specific issue this quarter. We upgraded the configuration of a specific piece of equipment that we were going to buy [indiscernible] some delivery slots that were towards the end of this year. As we upgraded to a richer configuration, it swapped out to delivery slots from the end of this year, to Q1 of next year. So just shifted a few hundred million dollars worth of CapEx from 2015 to 2016."
Capex really doesn't matter. What matters is that they get yields where they have to be so they can start HVM.
Amazing how spending is irrelevant now that Intel has fallen behind on that front. I seem to recall that Intel's lead was primarily attributed to spending.
14nm still has issues though attributable to cost. Capex dropping, volume dipping, and process advancements being pushed out are all related. You're delusional if you think otherwise.
Intel's lead is nonexistent in mobile despite the supposed process advantages. Is it the uarch or the fabs that are the problem? Given the capex trend, I'm inclined to believe it's both -- if not, then certainly in the future.
They're ER looks decent because of increased asps in declining market with respect to volume. That's not a healthy business plan
The issue is the uarch. What intel needs to do is build a soc with core IP instead of Atom IP. They never will because they want to protect their higher price items, which is a shame. A SOC with core IP, very limited PCIe (maybe 2-4 lanes) and a built in modem, etc would be an amazing smartphone soc.
If you listen to the conference call, there's a tacit admission that Intel is pulling out of the mobile market and looking more to IoT to bring in appreciable volume, albeit at a more reasonable time frame -- years, not months.
With respect to modem, Intel has historically had lots of problems producing modems on its fabs. Even today the majority of consumer-facing modems are built at TSMC and not on Intel's own foundries. And producing them is one thing, making them at reasonable costs (read: low costs and high volume intended for the mobile market) is an entirely separate problem that doesn't jive with their cost structure.
We have Core-M, but compare that to the A9X and then consider the form factor. Apple's A9X can maintain its CPU frequency almost indefinitely with no GPU load, whereas Intel's Core-M throttles aggressively under the same conditions at a higher TDP and better cooling. Then consider the performance gap; or rather, what gap there is left.
Intel doesn't rule the roost anymore. The capex spending dipping at a time it should be increasing at an accelerating rate is evidence enough, imo. The fab advantage may be here now (and only when compared to AMD), but going forward Intel will be in a lot of trouble.
There are no devices available with the A9X yet. The A9 is in the iPhone 6s devices and the A8X is in the Air 2, but we won't have any info on the A9X performance until the iPad Pro in November.
Although I do agree with the overall sentiment of your post. Apple's most recent chips have impressed me more than Intel's recently. It will be interesting to see how this trend continues.
This is looking very dim. There is, as far as we can see right now, very little future in Client Side Computing, i.e Desktop. We will see what happen if Larger size iPad Pro further disrupt it. But ASP continue to fall. Smartphone Tablet - Not happening. IoT - despite its hype, is much more of an extension to mobile rather then separate category. DataCenter - The only good thing intel have right now, ARM wont be, and i would argue may never be able to compete in this sector.
Intel cant rely on Datacetre forever, where is their next big move?
The article mentions ASP three times, all increases.
But the volume is going down faster. Remember that if computers start lasting twice as long between upgrade cycles, that's a 50% decrease in sales volume per year without the installed base shrinking. There are plenty of client computers around, they're just being upgraded less often.
I have a hard time understanding how ASP is increasing in Client Side Computing, it must mean Intel are pushing less cheap Pentium and selling out more Skylake.
Similar on the Server Side, Intel has nearly abandon the Atom Server, or mainly because there is no demand for it. Xeon-D is selling much better then expected.
ASP up in client computing makes sense I think? If people are buying/replacing their computers (much) less often then you will also get them a bit more willing to spend a premium in the first place.
With how the market is set up, that means Intel getting a fair chunk of it.
Most client PC's are laptops and chip costs in laptops doesn't seem to go down. Core-M chips have a list price of around 300 USD. Even cheap laptops tend to have higher end CPU's than they did a few years ago.
The Datacenter, whether cloud or not (doesn't matter to Intel) will continue to grow for a long time. We're producing a shitload of data as consumers, and it's being increased artificially all the time. You think you need 20MP on a cell phone camera? No. But people upload pics 'to the cloud' as they shoot them and they shoot many. Videos, same deal. It all needs more storage, faster networks, all the things Intel loves to provide.
@hansmuff Next big move? Integrating FPGA and 3DX into the CPU. And after 10 more year of that and miniaturization of the chip and improvements of the camera/video, holographic images, the begging of the Star Wars Saga.
<i>Intel cant rely on Datacetre forever...</i> Why not? If all desktops and laptops are replaced with smart phones and tablets, then data center will be extremely lucrative. Weak, thin-client devices like that are useless without data centers. In any case, the death of the pc is overblown. Reduction of the pc is more like it. Keyboard, mouse, and large monitor are still needed to get much work done.
I know there a plenty of people excited about the downfall of intel, but somehow I think they are still on a good thing. Definitely their are issues; such as the long term structural problems with moving to mobile everything, the SSD revolution, the need to push power efficiency, and smaller nodes costing more per chip.. Still, they are making tons of cash. I wouldn't be worried too much. They just need to improve Atom again and they will find more space in the low cost markets to find their volumes again. Otherwise they need to get a new uuarch which makes i3/i5/i7 on 5 watts more impressive at same cost. Two things they probably aren't interested in doing anyway.
Lastly, I would not be surprised if tsmc, etc all hit the same roadblock after 16nm
Intel.. to the eternal absolute truth is better than AMD. So is NVIDIA. AMD has to jack up power requirements just to keep up and people who have no idea but get tricked by good marketting go AMD. Despite AMD being the entire house Xbox One sold all over the world and some computers.. they only got a revenue of $1B while Intel got $14.5B. In the long run people spend more on the electric bill than going Intel during the initial ownership. Looking mid and long term is more important than the early term if you are smart.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
23 Comments
Back to Article
Shadow7037932 - Tuesday, October 13, 2015 - link
Ouch. 39% drop in tablet sales. I guess this is to be expected as tablets overall don't seem to be doing very hot right now.nofumble62 - Tuesday, October 13, 2015 - link
Actually it's a good news. The end of contra-revenuenathanddrews - Wednesday, October 14, 2015 - link
I, for one, will gladly continue to shop Intel's clearance rack.Morawka - Wednesday, October 14, 2015 - link
that's not year over year,, thats just the quarter.. Its not as bad as you might think.Intel's Performance cpu's are just now starting to use 5w of power... imagine in in a couple of years when 10nm is out and you have core i5 performance in a fanless 5w package. And the low power stuff can be used in phones at 2.5w
medi03 - Thursday, October 15, 2015 - link
And why on earth would anyone want that? Exactly what would Intel's chip in a phone do, what ARMs do not?mrdude - Tuesday, October 13, 2015 - link
Another dip in CAPEX at -$400million.This doesn't exactly exude confidence from a fab perspective or even longer term performance. Not when Samsung and TSMC are increasing spending and even exceeding Intel.
witeken - Wednesday, October 14, 2015 - link
The 400M drop is irrelevant:Stacy Smith: "Actually, it’s a pretty specific issue this quarter. We upgraded the configuration of a specific piece of equipment that we were going to buy [indiscernible] some delivery slots that were towards the end of this year. As we upgraded to a richer configuration, it swapped out to delivery slots from the end of this year, to Q1 of next year. So just shifted a few hundred million dollars worth of CapEx from 2015 to 2016."
Capex really doesn't matter. What matters is that they get yields where they have to be so they can start HVM.
mrdude - Wednesday, October 14, 2015 - link
Amazing how spending is irrelevant now that Intel has fallen behind on that front. I seem to recall that Intel's lead was primarily attributed to spending.14nm still has issues though attributable to cost. Capex dropping, volume dipping, and process advancements being pushed out are all related. You're delusional if you think otherwise.
Intel's lead is nonexistent in mobile despite the supposed process advantages. Is it the uarch or the fabs that are the problem? Given the capex trend, I'm inclined to believe it's both -- if not, then certainly in the future.
They're ER looks decent because of increased asps in declining market with respect to volume. That's not a healthy business plan
extide - Wednesday, October 14, 2015 - link
The issue is the uarch. What intel needs to do is build a soc with core IP instead of Atom IP. They never will because they want to protect their higher price items, which is a shame. A SOC with core IP, very limited PCIe (maybe 2-4 lanes) and a built in modem, etc would be an amazing smartphone soc.mrdude - Wednesday, October 14, 2015 - link
If you listen to the conference call, there's a tacit admission that Intel is pulling out of the mobile market and looking more to IoT to bring in appreciable volume, albeit at a more reasonable time frame -- years, not months.With respect to modem, Intel has historically had lots of problems producing modems on its fabs. Even today the majority of consumer-facing modems are built at TSMC and not on Intel's own foundries. And producing them is one thing, making them at reasonable costs (read: low costs and high volume intended for the mobile market) is an entirely separate problem that doesn't jive with their cost structure.
We have Core-M, but compare that to the A9X and then consider the form factor. Apple's A9X can maintain its CPU frequency almost indefinitely with no GPU load, whereas Intel's Core-M throttles aggressively under the same conditions at a higher TDP and better cooling. Then consider the performance gap; or rather, what gap there is left.
Intel doesn't rule the roost anymore. The capex spending dipping at a time it should be increasing at an accelerating rate is evidence enough, imo. The fab advantage may be here now (and only when compared to AMD), but going forward Intel will be in a lot of trouble.
cmikeh2 - Wednesday, October 14, 2015 - link
There are no devices available with the A9X yet. The A9 is in the iPhone 6s devices and the A8X is in the Air 2, but we won't have any info on the A9X performance until the iPad Pro in November.cmikeh2 - Wednesday, October 14, 2015 - link
Although I do agree with the overall sentiment of your post. Apple's most recent chips have impressed me more than Intel's recently. It will be interesting to see how this trend continues.Michael Bay - Thursday, October 15, 2015 - link
Fab advantage? Literally everybody will stay on 14/16 for a decade, except Intel, which will simply have to push lower nm.iwod - Tuesday, October 13, 2015 - link
This is looking very dim.There is, as far as we can see right now, very little future in Client Side Computing, i.e Desktop. We will see what happen if Larger size iPad Pro further disrupt it. But ASP continue to fall.
Smartphone Tablet - Not happening.
IoT - despite its hype, is much more of an extension to mobile rather then separate category.
DataCenter - The only good thing intel have right now, ARM wont be, and i would argue may never be able to compete in this sector.
Intel cant rely on Datacetre forever, where is their next big move?
stephenbrooks - Wednesday, October 14, 2015 - link
--[But ASP continue to fall.]--The article mentions ASP three times, all increases.
But the volume is going down faster. Remember that if computers start lasting twice as long between upgrade cycles, that's a 50% decrease in sales volume per year without the installed base shrinking. There are plenty of client computers around, they're just being upgraded less often.
iwod - Wednesday, October 14, 2015 - link
I need to learn to read properly ><I have a hard time understanding how ASP is increasing in Client Side Computing, it must mean Intel are pushing less cheap Pentium and selling out more Skylake.
Similar on the Server Side, Intel has nearly abandon the Atom Server, or mainly because there is no demand for it. Xeon-D is selling much better then expected.
Qwertilot - Wednesday, October 14, 2015 - link
ASP up in client computing makes sense I think? If people are buying/replacing their computers (much) less often then you will also get them a bit more willing to spend a premium in the first place.With how the market is set up, that means Intel getting a fair chunk of it.
Penti - Wednesday, October 14, 2015 - link
Most client PC's are laptops and chip costs in laptops doesn't seem to go down. Core-M chips have a list price of around 300 USD. Even cheap laptops tend to have higher end CPU's than they did a few years ago.hansmuff - Wednesday, October 14, 2015 - link
The Datacenter, whether cloud or not (doesn't matter to Intel) will continue to grow for a long time. We're producing a shitload of data as consumers, and it's being increased artificially all the time. You think you need 20MP on a cell phone camera? No. But people upload pics 'to the cloud' as they shoot them and they shoot many. Videos, same deal.It all needs more storage, faster networks, all the things Intel loves to provide.
Next big move? I'd be charging a lot if I knew.
Vlad_Da_Great - Thursday, October 15, 2015 - link
@hansmuff Next big move? Integrating FPGA and 3DX into the CPU. And after 10 more year of that and miniaturization of the chip and improvements of the camera/video, holographic images, the begging of the Star Wars Saga.Jaybus - Friday, October 23, 2015 - link
<i>Intel cant rely on Datacetre forever...</i>Why not? If all desktops and laptops are replaced with smart phones and tablets, then data center will be extremely lucrative. Weak, thin-client devices like that are useless without data centers. In any case, the death of the pc is overblown. Reduction of the pc is more like it. Keyboard, mouse, and large monitor are still needed to get much work done.
doggface - Wednesday, October 14, 2015 - link
I know there a plenty of people excited about the downfall of intel, but somehow I think they are still on a good thing. Definitely their are issues; such as the long term structural problems with moving to mobile everything, the SSD revolution, the need to push power efficiency, and smaller nodes costing more per chip.. Still, they are making tons of cash. I wouldn't be worried too much. They just need to improve Atom again and they will find more space in the low cost markets to find their volumes again. Otherwise they need to get a new uuarch which makes i3/i5/i7 on 5 watts more impressive at same cost. Two things they probably aren't interested in doing anyway.Lastly, I would not be surprised if tsmc, etc all hit the same roadblock after 16nm
AntDX316 - Friday, October 23, 2015 - link
Intel.. to the eternal absolute truth is better than AMD. So is NVIDIA. AMD has to jack up power requirements just to keep up and people who have no idea but get tricked by good marketting go AMD. Despite AMD being the entire house Xbox One sold all over the world and some computers.. they only got a revenue of $1B while Intel got $14.5B. In the long run people spend more on the electric bill than going Intel during the initial ownership. Looking mid and long term is more important than the early term if you are smart.http://www.anandtech.com/show/9172/amd-posts-q1-20...