The AMD Execution Thread [2007 - 2017]

[3dilettante]

I am not inplying mubadale should pay more or less. That way of looking at it is uninteresting in my world and you have lawyers and accountants for that.

Legal reasons are why the arrangement between AMD, GF, and Mubadala was so tortured.
AMD had to be the owner in some legal sense because it outsourcing the majority of its x86 manufacturing would allow Intel to challenge AMD's adherence to a cross-licensing deal that kept their respective products out of IP litigation.
Intel eventually challenged it anyway.

There is a reason we see a change to the wsa today. Longer period, fines for using tsmc/ss and inderectly mubadala ownership in amd.

The exact mix of problems is unclear. AMD is obligated to buy some amount of volume from GF, and that is already something that it is having problems doing because it is not very successful.
Further, it is potentially reducing its volume further by manufacturing products originally agreed to be built at GF at other foundries, likely due to GF being not very successful in this portion of its business.

AMD wants to make decent products that it can sell, which GF apparently is not facilitating well at this node or the 10nm node it is skipping.
GF has put aside engineering resources and a share of the capacity and overhead of multi-billion dollar facilities on specific variations of the tech that are generally important only to AMD.

What this agreement is actually doing is formalizing the regular penalties AMD has taken before for very similar situations, which were theoretically "temporary" in nature.

And why is it possible now and not back then if it was even considered?

AMD's agreement with Intel was changed to allow outsourced production.
The price, from what I can tell in a version of the agreement, is that it no longer covers AMD's processors after a deadline that has expired.
Zen's matching Haswell's base ISA and standard extensions may be a symptom of the end of AMD matching Intel's ISA changes going forward.

 

[Krumme]

As i underrand it a major part of the 1.3B 2009 settlement was that amd could outsource production. Within a few months after the deal amd announced bobcat at tsmc 40nm. And it then ran there full steam. No waiting. Thats 2009.

 

[3dilettante]

The agreement between Intel and AMD was amended in late 2009, the fab split was formalized earlier in the year and was disputed soon after.
The disputes over litigation over Intel's anti-competitive actions and the GF split's legitimacy were resolved the new agreement, a comparatively small payment for the alleged damages Intel inflicted, and an end to any further litigation from AMD.

Even with the older agreement, a minority share on the order of 20% could have been outsourced. Intel disputed whether AMD was outsourcing everything with the spin-off.

 

[Krumme]

Ok. So why does it take 7 years to alter the agreement?
From 2011 it was evident bd was failing.
And why does it have a longer period now and not longer before?

 

[3dilettante]

Ok. So why does it take 7 years to alter the agreement?

Which agreement are you referring to?
The WSA has been renegotiated regularly, and amendments and special charges like this one done before.
The original structure of the AMD/GF/Mubadala deal is no more. AMD sold off its stake after it got out of its manufacturing obligations.

From 2011 it was evident bd was failing.

I'm not sure what change of arrangement with GF or AMD's creditors would do to change that.
The BD core design and architectural philosophy were a misfire, and high-end architectural cycles are ~5 years, which Zen serves as an example of.

And why does it have a longer period now and not longer before?

That's something AMD and GF would know. It would remove the frequent periods of uncertainty and give a predictable accounting for GF's revenue and AMD's incurred costs. I think the preference would have been that AMD could have been able to meet its customary volume obligations and that GF could satisfy in terms of volume and process competitiveness.

A negative interpretation is that it is uncertain that AMD can generate the necessary demand, and that it is uncertain that staying with GF will allow more demand to be generated. Regardless, AMD is not allowed to leave for quite some time, even if GF looks worse for being paid to not be used.

 

[Krumme]

Thanx for the explanations.

I am totally obsessed myself in using fixed capacity near 100%. Things that have 6-10 years depreciation. The impact on profitability screams out loud.

No wonder its the first thing that comes to mind from mubadala when making the agreement bact then and the wsa rev from then on. Its basic risk control. Hard not to control your mind.

Still. I am trying to learn from this. I cant help to think if amd hadnt been better off dumping bd as arch day one and gf better off with a bag of money and a message to focus elsewhere in something with more perspective.

It seems to me the agreement kind of hold both in a position where they didnt take the nessesary decisions.

Probably the outcome have been more or less the same had they not split up but innovation on the business side seems to me have been hindered.

Look where gf got since 2009? Nowhere. At least amd seems to have developed some kind of niche. Its not a good story and both are still locked in this cave only escaping if the zen miracle should happen. And miracles dont happen.

 

[3dilettante]

Still. I am trying to learn from this. I cant help to think if amd hadnt been better off dumping bd as arch day one and gf better off with a bag of money and a message to focus elsewhere in something with more perspective.

Per this, Zen was started in late 2012, there is some turnaround after it was clear BD was unable to gain traction and a resources had to be shuffled. Zen is not likely to really get to market until next year, so a ramp and server deployment is 4+ years.
http://semiengineering.com/the-zen-of-processor-design/

For AMD's part, they still needed GF and would have negotiated some requirement that they not be left without any fab capacity. Bulldozer and Piledriver were wedded to the 32nm SOI process at GF, and the bulk 28nm CPUs had a process that was partly customized for AMD's needs.

With the 14nm generation, it's not clear how much of a barrier there is for the various products, although the new WSA with a commitment to 7nm sounds like AMD is not really escaping the need/obligation of a GF process.

It seems to me the agreement kind of hold both in a position where they didnt take the nessesary decisions.

The number of options was limited.

Probably the outcome have been more or less the same had they not split up but innovation on the business side seems to me have been hindered.

If they hadn't split, AMD would have imploded. The financial burden of process development and chronically underutilized fabs were beyond its ability to sustain, without taking into account a long list of problems ranging from mediocre products, process problems, capacity constraints followed by capacity glut, overpayment for ATI, problems incorporating ATI or realizing synergies with it, loss of staff, and inventory/channel management problems.

This AMD gloom thread is an impressive testament to how long a downward spiral corporations can exist in.

Look where gf got since 2009? Nowhere. At least amd seems to have developed some kind of niche. Its not a good story and both are still locked in this cave only escaping if the zen miracle should happen. And miracles dont happen.

Since 2009, GF absorbed Chartered Semiconductor and IBM's manufacturing arm. Financially, it hasn't look great, but it has business outside of AMD. Whether it's good enough for Mubadala when the oil income that finances these plays is threatened is unknown to me.

 

[Krumme]

Beeing no 2 in a market where there is natural monopoly is near impossible.

Without insight into this this specific business people would have been able to predict the outcome.

Amd have practically never earned money for like 35 years except a few years that was by purely by extreme chance. K7 from outside team. Ibm soi. P4 blunder. Lots of lipstick on results during the years after phenom up until even the numbers were red again. As they have always been.

But the criticism of amd is wrong and have always been. Its the laws of economics that prevent them for earning money when they try to do more or less the same as Intel.

Reducing the problems to eg constant lack of execution bad arch whatever fails to adress the most obvious. Trying to compete with Intel or nv in their turf is bound to fail. It simply cant be different. And its just a long road consistent leading down the drain. Same pattern in other similar business. Beeing the small guy is not an option unless you want to get beaten and beeing in constant trouble.

Now amd says they have left that pattern and granted some elements of zen arch point to thats. Smalller than the big cores fatter than small. We dont know how it pans out but this time there actually is a chance to earn money because they stay off the main path. But if we look at amd besides that its constantly the same problems. A wide portfolio in gfx where they are always second to nv and at the same time going for cpu server notebooks embedded you name it. How on earth is that a strategy. Its difficult to spread ressources more. So unfocused.

Shortsighted revenue focused tactics imo. Having the need to fill an utterly uncompettitive fab doesnt help but just pull the wrong direction.

 

[Deleted member 2197]

AMD Gaming Evolved APP By Raptr No Longer Being Supported By AMD

Starting September 12th 2016, AMD is no longer bundling the “AMD Gaming Evolved App” by Raptr with builds of Radeon Software. The application will still work. AMD will cease to undertake any compatibility testing, install support or general technical support for this application, nor will it be available through Radeon Software or its installer. Previous builds of Radeon Software that include the “AMD Gaming Evolved App” dated before September 12th 2016 will remain intact and will not be affected. -AMD

http://wccftech.com/amd-gaming-evolved-app-raptr-no-longer-supported/

 

[keritto]

Configuration Options: ....
• Configurable to 3.5GHz 6-core processor with
12MB L3 cache, 3.0GHz 8-core processor with
25MB L3 cache, or 2.7GHz 12-core processor with
30MB L3 cache
• Configurable to 16GB (four 4GB), 32GB (four
8GB), or 64GB (four 16GB)
• Configurable to dual AMD FirePro D500 with 3GB
of GDDR5 VRAM each or dual AMD FirePro D700
with 6GB of GDDR5 VRAM each
• Configurable to 512GB or 1TB PCIe-based flash
storage

I did wonder where did Apple get those fancy vaporware CPUs for a while. Its good that ark.intel is available to show what they currently produce

or was this just another hyperfalse advertising from ever stretching blob agency

Or does Apple has yet another under the desk deal (YAUDD) to pick up from exclusive Broadwell-E MAXXX 10+2core 25+5MB L3 ?

 

[RedVi]

I did wonder where did Apple get those fancy vaporware CPUs for a while. Its good that ark.intel is available to show what they currently produce

...

Or does Apple has yet another under the desk deal (YAUDD) to pick up from exclusive Broadwell-E MAXXX 10+2core 25+5MB L3 ?

Getting a bit OT here, but the CPU's apple use in Mac Pro's would be Xeon's.

 

[Dave Baumann]

http://247wallst.com/technology-3/2...rs-with-new-alibaba-collaboration/?yptr=yahoo

Alibaba adopting FirePro S7150 x2 for Datacenter cloud services.

 

[Deleted member 2197]

AMD reports $1.3 billion in Q3 revenue on higher console chip sales

AMD reported revenue of $1.3 billion (on a non-GAAP basis) in the third quarter ended September 30, compared with $1.06 billion a year earlier. Net profit on a non-GAAP basis was 3 cents a share, compared with a loss of 17 cents a share a year ago. Analysts had expected a non-GAAP loss of 1 cent a share on revenue of $1.21 billion.

The record semi-custom chip revenue and higher GPU revenue in the third quarter was partially offset by lower client desktop processor and chip set sales. Gross profit margin was 5 percent, down from 31 percent in the previous quarter due to a $340 million charge related to buying chips from contract manufacturer Globalfoundries. Expenses were higher in a part due to more investment in research and development.

AMD’s debt now stands at $1.6 billion, down $606 million from the previous quarter. AMD had previously guided full year revenue would increase in low single digits. Based on mid-point of our Q4 guidance, 2016 revenue guidance has been increased to 6 percent. AMD’s stock is down 1.8 percent to $6.83 a share in after-hours trading.

http://venturebeat.com/2016/10/20/a...n-in-q3-revenue-on-higher-console-chip-sales/

 

[homerdog]

How many quarters in a row can AMD lose hundreds of millions of dollars before they flatline? Can they continue indefinitely like this? I guess as long as people will invest in them they can float.

In any case AMD isn't even fulfilling intended role any more (to keep Intel in check). Intel does whatever it wants already. There is little to no pressure from AMD except in the budget range where there are little profits to be made. This is pretty much the best case scenario for Intel: no competition and no antitrust concerns.

 

[tunafish]

How many quarters in a row can AMD lose hundreds of millions of dollars before they flatline? Can they continue indefinitely like this? I guess as long as people will invest in them they can float.

Pretty much. Thanks to the stock boost, they could turn some debt into equity, and despite losing $400M, their balance sheet is now $900M healthier than it was last quarter. They lost almost half a billion, yet they are now almost a billion less in a hole. More than that, almost half of their remaining debt is now in a $800M convertible note with negligible interest. I have no idea why someone would buy that, yet someone did. AMD's CFO continues to just astound me.

AMD's balance sheet is now actually starting to look healthy, for the first time in a very long time.

 

[Razor1]

Well these are short term fixes, if they don't get product out to push their company's bottom line, future stock price will just go back down. What they did with the price stock jumps *selling common share to reduce debt* perfect. What they have done so far puts the company in a position for a buyout. But again if they don't have the product to back up the stock price, reducing the debt is only half the problem.

 

[seahawk]

Zen and Vega might just be the winner products they need.

 

[tunafish]

future stock price will just go back down.

This will hurt investors, not AMD. Once they get rid of the remaining 1 billion of conventional debt, they couldn't care less about the stock price. AMD was genuinely teetering on bankruptcy for a while. They have basically climbed out of that hole. Going forward, they might not be an awesome company to invest in, but it seems like it's now safe to say that they will actually be around.

 

[Razor1]

This will hurt investors, not AMD. Once they get rid of the remaining 1 billion of conventional debt, they couldn't care less about the stock price. AMD was genuinely teetering on bankruptcy for a while. They have basically climbed out of that hole. Going forward, they might not be an awesome company to invest in, but it seems like it's now safe to say that they will actually be around.

Well with out decent products they will still be making looses per Q so that debt they have almost absolved will come back quickly.

 

[Dave Baumann]

The quarter was a 3 cents per share profit on a non-GAAP basis. In other words, outside of one time charges revenue generated from the products is already profitable.