haha!!! because jpmorgan wrote it its true? I use to work there trust me when I tell you gokul hariharan (author) is one of the most clueless people on wall st!
again wrong. TSMC 20nm has lower cost than Intel 14nm FINFET. In fact its Intel who is having bigger yield problems with 14nm FINFET than TSMC with 20nm planar.
https://markets.jpmorgan.com/research/email/-kjegkq4/GPS-1336259-0
page 4
"We estimate that TSMCs cost structure at 28nm 20nm is 30% lower than Intel.While Intel may narrow this somewhat at 14nm, we expect the cost gap to widen again at 10nm."
With TSMC 16FF+ ramping to volume production early next year you can see for yourself a faceoff against TSMC and Intel product. Apple A9 vs Broadwell Core M. This one is gong to be a good contest especially since both are FINFET and at same 5w TDP.
You forget that SoFIA is aimed that the entry and low-end market. I'm not sure if Qualcomm will get its 200 and 400 series to 20nm any time soon. A quadcore Silvermont, even at 28nm, will blow away any 200, 400 and 600 series SoC from Qualcomm (those SoCs with tons of slow A7 or A53 cores). It will be about on par with Qualcomm's 805 (I don't know about graphics, though).btw Baytrail-T built at Intel 22nm FINFET will move to an inferior TSMC 28nm planar HPM process. That affects perf and perf/watt and competitive positioning against the hordes of ARM competitors like Qualcomm, Mediatek and others.
No. you are wrong. Intel does not have baseband integration till SOFIA launches. Qualcomm is the king of integration. Their integrated modems are the best in the industry. To give you a clue as to how heavily integrated a Qualcomm SOC is have a look at the 20nm Qualcomm Snapdragon 810 or 28nm Qualcomm S800.
http://www.anandtech.com/show/7925/...bit-socs-with-lte-category-67-support-in-2015
http://www.qualcomm.com/snapdragon/processors/800
Cherrytrail will need a discrete modem chip like Intel XMM 7260 and that increases costs. So Intel Cherrytrail has no chance against Qualcomm 810 as cost is one of the major factors in the mobile SOC market.
http://www.intel.com/newsroom/kits/mobileworld/2014/pdfs/7260_factsheet.pdf
I encourage you to watch Intel's presentation about 14nm costs at their investor meeting in November. 14nm cost/transistor, certainly compared to what the fabless companies like Qualcomm pay, will be very good. Also, in the Q2 conference call, Brian says their 14nm problems are all fixed.again wrong. TSMC 20nm has lower cost than Intel 14nm FINFET. In fact its Intel who is having bigger yield problems with 14nm FINFET than TSMC with 20nm planar.
https://markets.jpmorgan.com/research/email/-kjegkq4/GPS-1336259-0
page 4
"We estimate that TSMCs cost structure at 28nm 20nm is 30% lower than Intel.While Intel may narrow this somewhat at 14nm, we expect the cost gap to widen again at 10nm."
With TSMC 16FF+ ramping to volume production early next year you can see for yourself a faceoff against TSMC and Intel product. Apple A9 vs Broadwell Core M. This one is gong to be a good contest especially since both are FINFET and at same 5w TDP. btw Cherrytrail will have to compete against S810 for high end tablet design wins.
You forget that SoFIA is aimed that the entry and low-end market. I'm not sure if Qualcomm will get its 200 and 400 series to 20nm any time soon. A quadcore Silvermont, even at 28nm, will blow away any 200, 400 and 600 series SoC from Qualcomm (those SoCs with tons of slow A7 or A53 cores). It will be about on par with Qualcomm's 805 (I don't know about graphics, though).
Intel has no leadership in 2015 or 2106. For 2015 Cherrytrail vs S810 is a no contest for phones as S810 is highly integrated. In tablets lets see how Intel fares. For 2016 you are talking about fully integrated Intel 14nm Broxton against Qualcomm fully integrated custom ARMV8 core SOC manufactured at TSMC 16FF+ / Samsung 14 LPE. Again lets wait and see actual products.The only thing Intel lacks is the integrated modem, which they don't have because apparently it takes a long time to move from TSMC to 14nm. But if you've seen Intel's slides at MWC and Brian's comments at the earnings call yesterday, you can rest assured that Intel knows this and they aim for a leadership position in 2015-2016. But Qualcomm isn't doing any sort of integration wizardry that Intel can't do.
I encourage you to watch Intel's presentation about 14nm costs at their investor meeting in November. 14nm cost[/transistor, certainly compared to what the fabless companies like Qualcomm pay, will be very good. Also, in the Q2 conference call, Brian says their 14nm problems are all fixed.
What does the power deficiency compared to 22nm matter when 28HPM is still quite superior to 28LP? If Intel is willing to sacrifice margins to put SoFIA in the entry and low-end market where it's intended, that's nice. But the die size of SoFIA is very small (43mm²), so the cost should be good.Sorry but the low end is all about cost. Qualcomm snapdragon 410 and 610/615 are 28LP products while Baytrail is 28HPM. Qualcomm will continue to sell Krait based S800 products which will definitely do well against a TSMC 28HM Sofia. remember the transition to TSMC 28HPM comes at an power efficiency cost which means lower clocks compared to Baytrail-T as Intel 22nm FINFET is a superior process compared to TSMC 28HPM.
Intel has no leadership in 2015 or 2106. [...] Again lets wait and see actual products.
haha!!! because jpmorgan wrote it its true?
Intel has no leadership in 2015 or 2106. For 2015 Cherrytrail vs S810 is a no contest for phones as S810 is highly integrated. In tablets lets see how Intel fares. For 2016 you are talking about fully integrated Intel 14nm Broxton against Qualcomm fully integrated custom ARMV8 core SOC manufactured at TSMC 16FF+ / Samsung 14 LPE. Again lets wait and see actual products
Regardless of this quarter, intel's revenue is flat going on 3 years. Operating income is also flat for 3 years now. With the average person's real income falling, who is going to keep buying more product to push revenue past this 3 year plateau? The top 0.1% can only buy so many cpus...
If you had really read the docs you would know that JPM is a market maker on both Intel and Altera stocks.
A market maker is a bank that is given a privilege in theses stocks tradings by the stocks markets owners.
They must provide liquidity to theses stocks and as such they must be on the two sides of the tradings, in the first lines of bid and ask, as such they can short sell a stock while being naked, meaning that they do not need to borrow the stock to short sell it, so their gain are little but without any risks unless they deliberately create a trend by publishing deliberatly flawed datas that would allow them to earn a lot without risks given their market makers status, it would be of course a breach of the contract and they could then be sued by whoever lost money on theses stocks as well as by EURONEXT or the NASDAQ.
Maybe it's just me, but I'm not really interested in information that one stands to profit from.How dare you! Implying jpmorgan is a more reliable source than message board conjecture.... smh
Costs are going down though. The cost for "good enough computing" is coming down quickly, and it won't be long before you're able to get a low end Windows desktop for $200.Regardless of this quarter, intel's revenue is flat going on 3 years. Operating income is also flat for 3 years now. With the average person's real income falling, who is going to keep buying more product to push revenue past this 3 year plateau? The top 0.1% can only buy so many cpus...
Maybe it's just me, but I'm not really interested in information that one stands to profit from.
.
The number is nowhere near that close. It's also not hard to verify much of the information being posted here... but you can't verify JPMorgan's commentary, for instance. They're also hinging their analysis a lot on TSMC's claimed density numbers compared to Intel's, which have already proven to be false (to clarify, both charts in Intel's claim and TSMC's rebuttal have errors).Then proceed to dismiss 50% of the comments made on this very subforum.
The number is nowhere near that close. It's also not hard to verify much of the information being posted here... but you can't verify JPMorgan's commentary, for instance. They're also hinging their analysis a lot on TSMC's claimed density numbers compared to Intel's, which have already proven to be false (to clarify, both charts in Intel's claim and TSMC's rebuttal have errors).
Volume production of 16 nanometer is expected to begin in late 2015 and will be fast ramped up in 2016. The ecosystem for 16 nanometer designs is current and ready. A few years ago, in order to take advantage of special market opportunities, we chose to develop 20 SoC first and then quickly follow with 16 nanometer. We chose this sequence to maximize our market share in the 20 nanometer, 16 nanometer generation.
Lora Ho - CFO
We'll start mass production of 15 nanometer from third quarter next year, while it's bit too early to tell you the percentage of revenue contribution, but we believe it will be a single digit range okay.
They also mention that 10nm is a 2.2x density improvement over 16nm... which isn't enough to catch Intel like they claimed they would, since Intel will also have a ~2.2x density improvement, and will be slightly more dense at 14nm.JPMorgan blew its credibility when it was claiming a 1H 2015 ramp of 16 FinFET+ (per a note linked by raghu78).
And are we citing the wonderfully objective SOI consortium on that?Well we dont need JPM to tell us that Intels 14nm FF will be more expensive than TSMCs 20nm Planar.
Same goes for 28nm Planar vs 22nm FF.
Qualcomm is the king of integration. Their integrated modems are the best in the industry. To give you a clue as to how heavily integrated a Qualcomm SOC is have a look at the 20nm Qualcomm Snapdragon 810 or 28nm Qualcomm S800.
Wow, did intel lose money last quarter? Sure seems like it from the tone of most of these comments. Talk about focusing on the negative. This thread went south in a hurry.
You wouldn't expect that from a forum that's generally regarded as being Intel biased, would you?Wow, did intel lose money last quarter? Sure seems like it from the tone of most of these comments. Talk about focusing on the negative. This thread went south in a hurry.
And are we citing the wonderfully objective SOI consortium on that?
You wouldn't expect that from a forum that's generally regarded as being Intel biased, would you?
Most of the concerns people have been expressing in this thread are borne out of ignorance. You simply can't compete with companies like Qualcomm without burning billions. It can be scary to see in the short term, as it takes years for those investments to even begin making a return, but eventually projects will complete and products will be delivered. It's at that point that judgement should be made -- once those products arrive. For all the flack Intel is getting, their stock price is going through the roof -- it seems that reality is what has the bias towards Intel, and members here are either living in the real world, or a dream world.
How dare you! Implying jpmorgan is a more reliable source than message board conjecture.... smh
Thread summary: Intel is doomed. Anyone who thinks otherwise is a fanboy.
Did I miss anything?
Thread summary: Intel is doomed. Anyone who thinks otherwise is a fanboy.
Did I miss anything?