MSFT approached AMD to see if AMD would buy MSFT's chip design business.
In response to a gaming Web site’s speculation Monday that Microsoft might try to buy chip maker Advanced Micro Devices, which supplies Microsoft’s Xbox game machine with its main processor, Bernstein Research analysts Stacy Rasgon and Mark Moerdler this morning argue the analysis makes no sense and there’s no good reason Microsoft would do so.
Rasgon, who has an Underperform rating on AMD shares, and Moerdler, who rates Microsoft Outperform, write that “investors have been asking if Microsoft would purchase AMD given recent ‘news’ articles,” but conclude the analysis, by KitGuru‘s Anton Shilov, “is not well thought out.”
"This is the latest rumor. It would be a major blow to Intel if true"
Not really. MSFT would try the purchase if they thought they could buy AMD cheaper than the price they have to pay for their XBOX parts. If this week's XXX is going to buy AMD rumor is true, it would not be any problem for Intel.
No one really wants to use AMD CPU parts. Even AMD is using Intel chips in their Project Quantum. That signals how good the AMD CPUs are.
If this prospect worries you, then you probably should reduce your INTC exposure and move back into safer investments.
A March 16th article in Fudzilla forecast what Digitimes announced now 3 months later. Duuuhhhh!
I would be more impressed with Ashraf if he was able to add something more than Digitimes rumor aggregation.
Intel putting finishing touches on SoFIA phones
by Fuad Abazovic on 16 March 2015
"The SoFIA SoC will get a 4G version later this year but we don’t think that 4G matters too much in this market segment as mobile network providers are simply charging too much for 4G plans compared to 3G and we don’t see many people with entry level to mainstream phones to end up paying for these expensive, unlimited 4G plans anytime soon. Those who choose such plans are much more likely to get a flagship phone on a two-year contract anyway.
We strongly believe that Motorola helped change the market with its affordable Moto G and Moto E phones, by offering decent performance at a very inexpensive price point. Since more and more phones are coming unsubsidized trough many providers in Europe, we believe that the price plays a major role in the phone decision making."
"but how could Skylake do this without anyone knowing about it by now? "
I don't think that SkyLake will integrate NVM but their are changes in the OS and hardware that will that will need to be made to properly support NVM memory. IMO, integration will happen at the low end first: mobile, Chromebooks, ....
hardware support of NVM memory
and there are a couple of U of Wisc papers that give a flavor of some of the things to consider when incorporating NVM memory.
and look at the dates of the recent changes. They are all this month. What would you think the library libnvdimm is providing? 8-)
II, there are too many moving parts to accurately assess. That is why every guesses at the numbers.
My guess is that Intel will be inside the ranges they gave at the start of the quarter, otherwise they would have "warned" and adjusted the limits up/down about a month ago. If they are outside the range, that would mean they did not know a month ago or they chose not to warn or orders collapsed .... all would be bad.
EVEN IF Intel meets their mid-points and affirm their high full year numbers, a number of the analysts will just assume that Intel is "lying" or is "clueless" and they will push their negative analysis.
I think this stock behavior will continue as long as GS Covello maintains his extremely negative position on Intel and has to manufacture the negative stories.
"(1) Micron announced poor results and weak forecast (most take it to indicate a poor PC market)"
The MU DRAM ASP was down 10% QoQ and that was the dominate impact on their revenues and profits. The ASP drop comes right out of the bottom line. Their costs were a little higher per bit which made the problem worse. Their "bits" total was fairly flat. They also have a staggered quarter that ends 2 months later than Intel so they overlapped the bad Intel Q1.
"2) With less than a week left in Q2, Intel hasn't pre-announced (so its results may be in line with previous guidance)"
That just means that the Intel results are in their range but the reaction will be negative if below the mid-point. The market is already expecting a lowered forecast from Intel, IMO.
"(1) Weaker results for DRAM manufacturers like Micron"
The MU results were from a drop in ASP which indicates weak demand.
"2) Higher Skylake ASPs resulting in increased revenues for Intel"
The Purple Swan article quotes an Intel patent as indicating no DRAM in mobile so that will raise mobile ASP to include mobile DRAM, NAND, DRAM controller and packaging/build costs into the Intel device. Down in the Purple Swan, Breezy talks about packaging being discussed for Knights Landing and the die are all arranged in a single package with photonics interconnects. Single package systems. SiP .... System in a Package.
"3) Compelling reason for Intel to expedite Skylake at the expense of Broadwell."
This is the only reason that I can think of .... that their is something big with the memory architecture to drive down power and dramatically increase value (and therefore ASP).
ALL IS JUST SPECULATIONS BUT IT SOUNDS GOOD. Intel has never been able to keep a secret like this.
"maybe someone from SA could look at this"
Good idea ....
Intel And Micron: The Purple Swan
Jun. 11, 2015
Breezy's hobby is reading patent applications and he found a couple of Intel/Micron patents talking about DRAM replacement. This is a KEYSTONE article that chains together some history and technical details that is possible to plow through at a number of levels.
... AND .....
It will all play together in the Xeon SkyLake Purley Server platform.
Intel Corporation's Purley Server Platform Looks Like a Beast
May 26, 2015
"In one of the slides, Intel says Purley is the "biggest platform enhancement since Nehalem." "
"Intel says it will be bringing "all-new memory architecture" that will enable four times the capacity relative to its last platform. It also claims to allow for "persistent data" (the contents of memory don't go away when power is removed), but at the same time is 500 times faster than traditional NAND flash."
"Why would you say that data center ASPs were also down"
I think your hearing is fine.
MU said bit volumes were down 2% but their 10% drop in ASP and increased costs affected their revenue/profits most. I was reading MU slide 15.
The MU results discuss "PC" weakness and my comments were limited to the PC volumes in Q1, not data center. Intel saw weakness in their fiscal Q1 (Jan/Feb/Mar) and MU saw weakness in their Q3 (Mar/Apr/May).
"Bellweather for PC memory ?"
Possibly. MU DRAM ASP's were down by 10% in the quarter, their costs were up slightly and number of bits down by 2%.
Down volume of 2% is about what Intel saw in Q1 too wasn't it?
"MU eanings miss"
MU down 12% after hours on 7.5mil shares.
INTC down 0.8% in sympathy on 200k shares.
If INTC is getting "crushed", I wonder what is happening to MU which is down 15x INTC ??
This looks rather nice. You can replace your motherboard with the Mini-ITX form factor board from Supermicro for $789.99 INCLUDING CPU. You will probably need some DDR4 memory but can re-use your peripherals and get an 8-core system.
About $200 per 8gb of DDR4 memory now.
Supermicro X10SDV-F-O Intel Xeon D-1540/ DDR4/ SATA3&USB3.0/ 2GbE/ Mini-ITX Motherboard & CPU Combo
CPU (Included): Intel Xeon D-1540 Processor; Support up to 45W TDP (Eight-Core)
Chipset: System on Chip
Memory: 4x DDR4-2133/ 1600/ 1333MHz DIMM Slots, Support up to 128GB ECC/REG Memory or up to 64GB ECC/Non-ECC Unbuffered Memory
"No down thumbs"
I thought it was a very interesting find. The Intel Devil's Canyon was put in very early models which makes me think that AMD did not even try to avoid detection by shipping AMD versions first.
"Apple move to make bitcode mandatory for iOS apps "
An Apple move to bitcode would ALSO reduce the risk of Apple systems being infected by viruses and other nasty things that native binary code can do. It could simply be a SECURITY move on Apple's part.
It might be interesting to compare the bitcode format with the Intel Houdini library to see the distance between the ARM instruction set vs the instructions that Houdini would have most problems with.
"Alex--If you are asking me about the source for the Canadian analysis, there are several for the migration of Canadians to the U.S."
That is exactly what I was asking for. Thanks much.
I especially like that the Commonwealth Fund has a non-partisan structure. Frasier Institute seems to lean conservative/libertarian. From the first Wikipedia paragraph on each. I like it when the underlying numbers from different organizations line up. I tend to give that data more weight.
The Wikipedia page indicated the Canadian Heath Survey included 17k of their "members" and found that 0.5% of them had used the US health care system. Of that 0.5% less than 1/4 said they went across the border specifically for the US health care Extrapolating the 0.5% of all the Canadians, a 42k number sounds about right and might even sound a little low. It is probably not possible to figure out reason for the Canadians being in the US (business, vacation or healthcare) at the time.
I was trying to understand the Canadian issue a little more and try to figure out the historical number and trends. It seems like the Canadian number includes all Canadians entering the US health system. That includes those who want to and those who have no choice:
1. Canadian tourists who have health emergencies.
2. Canadian businessmen who have health emergencies.
3. Canadians who intentionally seek heath care at a US installation.
It is hard to separate out the 3 groups from the raw statistics and I assume that all 3 groups are included in the totals cited. Even those in the first 2 groups who do not want US supplied care would be included in the "traveled to the US for care" total.
I always like to start with an UNCHALLENGED Wikipedia page because I think it contains information devoid of partisan tilt ... just an assumption I hold. Do you have a good source for the Canadian analysis?
The Wikipedia page on Healthcare in Canada has a section on visiting Canadians.
Healthcare in Canada : Canadians visiting the US to receive health care
"How many times do I have to say this?"
I wish it had been zero.
"There isn't enough information to make an estimate"
"Now you are suggesting that you had additional information and didn't present it. "
NO. I am suggesting that you do not know if I have additional information.
"Which is even more unethical."
Silly hyperbole. The post was ethical in either case.
"I'm just saying that a prudent individual would have skipped trying to make an estimate"
A prudent individual could have gained some information from my posted equation. The responsible individual would have been able to disagree without the attempted bullying.
The prudent individual is a careful investor who would try to make the estimate and size the impact on his company.
"Your assumptions are connected to absolutely nothing"
There was nothing wrong with my statement (copied here). Rather than bully, insult and call people names, you COULD calmly suggest your estimates and share ideas. Your choices define you.
"If a "few hundred" is 500 and there are 20 Intel sites, it could add up to a larger number like 10,000."
"You simply don't have enough information to make a prediction. And yet you do so anyway."
Of course, you do not have enough information to make THIS prediction and yet you do so anyway. You have no idea what information I actually have access to. You know what is in the public domain AND you ASSUME that is the ONLY information I have. You do not know what information I have access to. Think about this carefully .... and you my get a chuckle (or really frustrated) at the irony of your using the same technique you condemn.
" I'm disappointed to see this flawed methodology coming from you."
Then you should be frequently disappointed in yourself for using the exact same logic.
"You used a few hundred at EVERY location"
That is NOT what I did.
I SAID: If you interpret BK's non-specific statement with THESE ASSUMPTIONS AND I used the word "IF" to reinforce the possibility aspect (notice the word "IF" at the front of the sentence):
"If a "few hundred" is 500 and there are 20 Intel sites" ....
THEN the number could be 10,000.
The Intel LOCATIONS page identifies 64 "LOCATIONS" world wide. Some of the "Locations" like Santa Clara, Folsom, Chandler, Hillsboro identify areas with dozens of buildings in multiple cities with 1,000's of employees
SO WHAT I SAID was "math" and not "magic":
( 5 * 100 ) * ( 1/3 of 64 locations) = 10,000
"If this were a college quiz you would have gotten an "F".] "
It was a "Read Comprehension" quiz. I got the "A" and you got the "F".
"[Great, now you are doing magic numbers.] "
The BK quote you quoted and posted above said it was no more than a few hundred at each site.
The MATH is not magic. If you used your estimates for "few" and "sites" you can come up with the total you like. I made assumptions, disclosed them and performed the math. It was my estimate based on my assumptions.
( few x 100 ) * ( sites ) = layoff total
There are 64 actual Intel SITES according to the Intel Global Locations web page.
Intel global locations
The number was a few hundred per site. How many "sites" does Intel have? That would be a multiplier on the "few hundred". Did the "few hundred" just refer to the low performers or to all 3 category of employees laid off?
If a "few hundred" is 500 and there are 20 Intel sites, it could add up to a larger number like 10,000.