Typically speculators who are bearish on Advanced Micro Devices (NYSE:AMD) focus on the PC space, because that is where it is the easiest to make the bear thesis seen.
Sanford C. Bernstein's Dr. Stacy Rasgon did it not once, but twice, after the most recent earnings call. He even referred to AMD as a "hedge fund hotel." Man, that's harsh. Oppenheimer recently followed suit.
A common theme in the bear camp is that the PC market is in decline. In his most recent article, Seeking Alpha contributor Ashraf Eassa stated he would be interested in reading a counter-argument to the "PC" thesis, so I submit this article as a friendly rebuttal.
I Don't Fundamentally Disagree With What He Says
In his article, Ashraf argues that essentially Intel (NASDAQ:INTC) has better processors and will make it hard for AMD. This has been the story for several years, and more so after the release of Bulldozer in 2011.
Anyone invested in AMD that doesn't realize Intel has better big core processors really needs to do some due diligence.
But without looking at any sort of impact, it's an incomplete argument. Processor benchmarks do not mean anything without a discussion of market share.
I have explained before that AMD actually gained market share, in both desktops and notebooks, during Q2. During the recent Raymond James technology conference, AMD VP Mr. John Byrne explained that this notebook market share increase was driven via a 1.5M unit commercial shipment of AMD/HP (NYSE:HPQ) laptops sold in India.
No one said anything over this increased market share during Q2, but during Q3 when AMD was thought to have lost around 80 bps of market share, analysts jumped on this. The article on Benzinga where I am getting the 80 bps data from has an interesting quote, one which will serve to set the tone of my counter-argument:
But with PC revenues down an estimated 7.6% (vs. Intel up 3.5%), AMD lost 80bps of PC related market share, offsetting gaming.
I must've failed third grade math, because it is hard for me to figure out how a $50M decline in PC revenues offsets ~$400M in revenues from consoles. I'm still scratching my head.
And as an AMD investor, this is the frustrating part. Even if the opinions do not match reality, everyone is still entitled to their own opinion. This is what makes it so easy for the sell-side to focus on PC market share.
How Much Would PC Market Share Have To Drop To Offset Console Effects?
AMD's CFO, Mr. Devinder Kumar, pegged console operating margins at the "mid teen" level during the Q3 conference call, and AMD was believed to have generated ~$400M in console revenues. This equates to ~$60M in income at the operating level.
AMD has also been in the process of restructuring the company in order to trim OPEX to a more sustainable level. CS (computing solutions) generated ~$790M in revenues during Q3 with an operating income of $22M. Using typical corporate GMs of 40% (closer to what GMs would be without console sales since I am discussing CS revenue) would imply an OPEX for CS specifically of ~$294M, which I am rounding to $300M.
Again, using 40% GMs, AMD's CS revenues would have to fall to ~$600M for a decline in PC sales to offset console revenues from current levels.
Calculation: $600M * 40% GM = $240M in GM dollars
$240M - $300M in CS OPEX = ~-$60M, the amount needed to offset console sales during the most recent quarter
Note, these numbers are based on my assumptions above, but they should be illustrative of roughly how far PC sales would have to decline to actually offset console sales provided console sales of ~$400M. The amount of offset would change if console sales change.
A Brief Overview of AMD's Future Strategy in the PC Space
I am unconvinced of Bay Trail's superiority. First, I have made the argument before that Intel going after the low end space may not necessarily be in its best interest. I have been arguing since this summer that Bay Trail would not be good for margins if the intention is to replace Core sales with Bay Trail to stem PC losses. It's more a measure of stopping the bleeding rather than growing revenues.
Each chip made can behave slightly different. Some could consume a little more power, some could have CPU cores that don't work, etc. Intel is selling the best quad core Bay Trail parts at the lowest prices to power its way into tablets.
When you look at the desktop parts, the comparison is not nearly as drastic. Tom's Hardware managed to get their hands on a desktop Bay Trail part that Intel will be using to challenge the low end of the PC space. There will eventually be quad core Bay Trail Pentiums and Celerons available, but I have seen no extensive benchmarks of these chips yet, so the dual core part THG is testing is the most detailed I believe so far. If you would like to get a feel for actual performance, the Z3770 tablet chip benchmarks would be applicable as well for overall performance, but note the wattage will likely be higher on the notebook and desktop Bay Trail chips.
While I have provided the link for your reference, the one benchmark I will look at on THG shows that Intel and AMD measure TDP differently. AMD's 15W chip is on parity with an Intel 10W chip for power consumption. I'll also point out that when discussing wattage, small changes matter far more in tablets than in notebooks or desktops.
Intel's process advantage matters most in tablets, where the top binned parts with the lowest power consumption can be competitive. But in low-end desktops and notebooks, the extra couple of watts likely aren't going to be deal breakers.
So, onto how AMD intends to protect its hide in the PC space, a collection of leaks between articles on SWEClockers and PROHardver (as summarized on CPU World) give a good outline as to AMD's future plans.
The above graphs show where AMD is actively targeting different points within Intel's product stack in the sub $100 range.
Intel's top end mobile Bay Trail Pentium is the N3520, but it actually scores less in GeekBench than the tablet based Z3770 parts.
So don't think I'm being unfair in my benchmarks by comparing a tablet chip to a non-tablet chip, I'm actually using the higher quad core Bay Trail benchmark score.
Now I typically won't rely on leaks very heavily unless I can find a second data point for verification. Every now and then I get lucky and uncover a gem, and digging through GeekBench scores a while back I found this little beaut:
So here is the second data point I am referring to: the GeekBench 3 result I have provided above points to a 2.05 GHz Athlon 5350 socketed chip, in line with the leaked info.
The top tiered Athlon processor looks like it will be close to competitive with a ULV i3 part. Note that I think the i3 would likely beat the Athlon in many real world usage scenarios, but I think overall performance would be fairly close, and will use this to demonstrate a point in the conclusion.
When I look at actual performance of the quad core Bay Trail Pentium parts, Athlon looks to handily beat it in CPU performance. And the above result is believable, because it's essentially no different than a 2.0 GHz Kabini part.
AMD will also be releasing Beema and Mullins. These new chips should offer much improved performance/watt by implementing features like aggressive boosts.
Beema and Mullins are initially targeted at notebooks and tablets, and could help arrest any market share loss to Bay Trail in mobile. The socketed Kabini chips and aggressively priced Richland chips will help protect losses at the low end of the desktop space.
I don't have a crystal ball and will not attempt to take a stab at how much PC market share Intel could likely take in Q4 or beyond. In no way does Intel forcing its way into the tablet market using contra-revenues help AMD. But a drop from the current levels of CS revenues to $600M would represent a ~25% sequential decline, which seems highly unlikely.
The reality for AMD bulls is that the picture can essentially be painted anyway analysts want to describe it, and any loss of PC market share gives analysts the ability to continue this emphasis. AMD has very little market share in servers, and the semi-custom business is just getting started. The trick will be to see these revenues grow faster than PC market revenues contract. If this happens, the bear argument will continue to make less and less sense.
The reality for AMD bears is that AMD's engineers and management team haven't simply thrown their collective hands in the air and said, "Well we had a good run, but Intel is now gunning for more market share so it's time to punch the 'ole clock."
If the pricing information above in the leaks is close to accurate, AMD is offering competitive products at very low prices. The desktop Pentium is listed at $94, almost 2x the price of the Sempron at lower performance. And given that Jaguar is just over 100 mm^2, whereas Richland chips are 246 mm^2, AMD has more wiggle room on pricing when looking at die size.
There is also some notion among INTC bulls that if Intel targets the low end of the PC space that AMD will be the only casualty. This goes back to the benchmark of the Athlon creeping into Core i3 ULV territory. There is no magical binary area in which Intel can make the performance just good enough to only eat AMD chips and not cannibalize Intel silicon. If Intel prices and clocks the Bay Trail chips too aggressively, the company runs the real risk of eating into their own higher margin parts, which would at least partially offset the intentions of gaining market share at the low end.
This is not to say Intel will not try.
This whole endeavor in the near term will be a wait-and-see. If the leaks are accurate, AMD looks to have a good plan in place to protect market share. If you are long AMD, tracking PC market share and CS revenues for the next few quarters will be extremely important, as negative trends will feed the bears.
Author's Note: Document contains leaked information which is subject to change.
Additional disclosure: I own both shares and options in AMD and actively trade my position. I may add or liquidate shares or options at anytime, or I may initiate a small hedge via puts at anytime.