As weird as this may be, most of my articles on Seeking Alpha focus on stocks from a technology/product standpoint. Prior to Advanced Micro Devices' (NYSE:AMD) earnings call on July 18th, I would like to share my overall thoughts as to the company's financials for the year ahead. There is a lot to be said, so grab some coffee before you dive in. The page numbers in parentheses at various points throughout the article correspond to the pages from the 2012 annual report from which I am deriving much of my information.
Setting the Stage
For those (like me) that jumped on this bandwagon recently, I would like to look at 2012's financials to see how 2012 ended so badly.
Source: AMD IR Website
2012 Gross Margins
The first thing I would like to address is AMD's gross margin drop between 2011 and 2012. In 2011, AMD had a respectable GM of 45%, declining sequentially to 23% in 2012.
The decline in gross margins is explained away mostly in payments made to Global Foundries. $703M was agreed to be paid in order to break exclusivity and arrange pricing. The $703M was broken into a $425M cash payment recorded in 2012 and transfer of $278M of GloFo stock AMD owned back to GloFo. The second hit of $273M was for the third amendment to the Wafer Supply Agreement, reducing AMD's purchase requirements from GloFo (pages 41 and 45). There was also an inventory write-down of $100M during Q3 (page 50).
This is most important paragraph. According to various sources of information (I, II, III, IV), AMD had committed to purchase $900M worth of wafers in 2011 and $1.5B in 2012. Last year AMD entered into 2 separate WSA (wafer supply agreement) amendments. The first amendment served to break exclusivity from GloFo and address pricing. AMD shifted from good die pricing to take-or-pay, which could allow for positive surprise as yield improved. The second amendment deserves much focus. The overall impact of this amendment was 2 fold. The negative part, which received much attention, was AMD essentially paid GloFo to do nothing. AMD still had roughly $500M outstanding in required purchases to GloFo, and predicted that it would spend approximately $115M that quarter. In order to break the contract, AMD agreed to pay a termination fee of $320M, resulting in a net savings of $65M during FY 2012. The part that I would like to point out is that it also reduced the amount by which AMD was required to purchase from GloFo in 2013 by $350M (From $1.5B to $1.15B). And assuming AMD had initially agreed to $1.5B in 2014 (this is my assumption here), AMD agreed to purchase $250M in wafers Q1 2014, and would negotiate the rest of the pricing for 2014 sometime this year. Assuming that AMD was required to originally purchase $375M per quarter ($1.5B/yr), this results in another $125M in potential savings. Also, again my speculation, AMD may also be aiming to reduce purchasing commitments for 2014 to $1B ($250Mx4). So I believe the net effect is an overall savings closer to $220M ($65M Q4 2012 + $350M 2013 + $125M Q1 2014 - $320M fee). This number could be higher or lower though based on specifics of these amendments which we do not have access to as investors.
From reading the annual report, it appears these charges were all factored in for AMD's 2012 financials, but AMD still has 2 payments outstanding to GloFo. Given that these payments were charged in their entirety to 2012, the resulting physical payments that are made to GloFo during Q2 this year and Q1 next year should impact cash balance but not operating results.
Source: AMD Q1 2013 CFO Commentary
According to Mr. Kumar's comments, we can expect a $40M payment to GloFo to be announced during this earnings call, as well as another $200M Q1 next year. Given the sheer size of the fees for the amendments to the WSA, we can see why the associated write-downs knocked gross margin into the basement.
To start, we need to define the three groups that make up AMD's 3 operating segments: Computing solutions, graphics, and other. Computing solutions includes CPUs, APUs (CPU+GPU in the same die), chipsets, server processors, and embedded processors. Graphics include discrete GPUs for both mobile and desktop, workstation class cards, and game consoles (I have bolded these two as a point of discussion for later).
AMD's "All Others" category is mainly a non-reportable category which is used to figure in credits or expenses which are not from the sale of products, such as the write-downs I mentioned earlier as part of the amended wafer agreements between AMD and GloFo (page 47).
Computing solutions revenue saw a $1B decrease YoY. Approximately 85% of AMD's revenue is derived from legacy PCs, and AMD does not assume PC sales will pick up 1H2013 (page 24). The $1B drop was attributed to 14% decrease in shipments coupled with a 7% decrease in ASP of products. Note in 2011 revenue actually went up partly due to increased shipments from the Brazos platform, but the rise was partially offset by the lower ASP of Brazos CPUs (page 48).
Graphics saw a 9% drop in revenue during 2012, equating to a decrease in $150M. The reason cited was a 16% decrease in revenue from GPU sales. However, operating income went up $50M due to lowering expenses (page 49).
AMD lost revenue in 2012 from declining sales, accounting for a 17% drop YoY. This 17% drop was the result of a 20% drop in revenue of the computing solutions group and a 9% drop in the graphics division, resulting in a total revenue decline of $1.2B. Add this on top of the $976M paid to GloFo among other things and we arrive at a loss for 2012. The total loss for 2012 was $1.2B. Had AMD not had these hefty payments, net loss for 2012 would have been closer to $200M.
Also noteworthy is a $50M penalty incurred during the quarter from the sale and leaseback of property.
Computing Solutions Group is AMD's biggest revenue driver, and PC sales are expected to decline this quarter again.
This is the second most important paragraph. Last quarter PC sales declined 13.9% from the corresponding quarter a year prior. This quarter sales are expected to decline 11.9% compared to Q2 2012. There is a subtlety here that is very important to understand, and PC World explains it very well (and is the source of information, along with IDC). I highly recommend reading the articles by PC World linked to in this paragraph. According to IDC, 89.8M PCs were shipped during Q4 2012. So based on this number and the numbers from PC World for Q1 2013, PC sales declined roughly 15% from Q4 2012 to Q1 2013. IDC expects annual shipments to drop 7.8% in 2013, while Gartner expects a 10.6% decline. So assuming a 10% drop between 2012 and 2013, given a starting point of 352M based on 2012 shipments (techspot.com), this means roughly 316M PCs should ship this year. Given the numbers in the graphic below, this means there are still 165M PCs to be shipped during 2H 2013. If you assume both IDC and Gartner are wrong, and PC sales remain flat between each quarter in 2013 at 75M units shipped, this brings us to 300M units in 2013, and a drop of 15% from 2012 to 2013. If you're an optimist and assume the 10% drop is accurate, this means 82.5M units ship on average between Q3 and Q4.
Source: PCWorld.com, IDC
Most importantly, based on the numbers PC World provided, we can see that PC sales actually look fairly flat between Q1 and Q2.
AMD's Computing Solutions group saw a loss of $78M for a 9% overall revenue decline QoQ. AMD expected a rocky 1H2013 for PC sales according to their 2012 annual report. AMD has launched Kabini/Temash, which are updated versions of their popular Brazos platform. These low power chips are targeted at embedded systems, consumer products, and servers. AMD also updated their Trinity line of APUs with launch of Richland. Given the product launches, no revised guidance, and a possible slowing decline of PC sales, I assume revenue is essentially flat from the previous quarter at $750M.
AMD's graphics segment revenue rose 3% last quarter due to workstation cards, but operating income dropped $6M due to lower GPU sales to OEMs. I assume console graphics royalties take a small hit from lower sales due to impending refreshes.
Intel's (NASDAQ:INTC) Haswell iGPUs most likely exerted added pressure on AMD's entry level discrete GPUs. AMD has had no major product refreshes this quarter with the exception of the Radeon 8000M, while rival Nvidia (NASDAQ:NVDA) launched their 700 series. Based on this and an impending refresh of AMD's GPU lineup, I expect consumer desktop and mobile GPUs to take another small hit this quarter.
In regards to workstation class cards, AMD gained traction last quarter. AMD uses an open standards implementation, OpenCL, for coding programs which use their workstation cards, while rival Nvidia uses the proprietary CUDA implementation (Nvidia does however support OpenCL). I feel this gives AMD an advantage based on the fact that Intel's Xeon PHI accelerators also support OpenCL, meaning AMD's more powerful rival is also trying to shift the HPC market, which could benefit AMD. AMD also has announced design wins in Apple's (NASDAQ:AAPL) Mac Pro. Apple's Mac Pro appears to include two W9000 workstation cards. These cards retail for $3400 each. The Mac Pro will most likely be a low volume part, and AMD (my speculation) most likely offered these cards to Apple at a steep discount to Apple. Apple appears to have designed in house cooling for these GPUs, which will lower cost of production for AMD. To give an idea to the magnitude of this win, assume AMD sales these chips for $1250 each, totaling $2500 per Mac. Since the Mac Pro will most likely be low volume, assume 10,000 units are sold this year. This adds $12.5M in what is most likely high margin revenue. However, with the Mac Pro launching later this year, it will most likely not contribute directly to revenue this quarter, but possibly from the press received over the Mac Pro's announcement. I suspect AMD could have seen another revenue increase from workstation cards.
Overall, I assume revenue for the graphics segment comes in at $330M.
Given my estimations of $750M for Computing Solutions and $330M for Graphics solutions, we arrive at $1.08B in revenue total for Q2 2013, which is in line with management's guidance of $1.07-$1.11B.
I believe this is highly feasible provided PC sales remain flat between quarters. Another potential positive catalyst I mentioned in a previous article is management's drive to lower expenditures. There is potential for roughly another $10M in savings between this quarter and last as AMD aims to reduce OPEX to $450M by Q3.
Based on the fact I believe that AMD will come in at guidance, the most important part for Thursday I feel will be guidance.
This is the third most important paragraph. The biggest factor I have not addressed yet is console wins. There has been some misconception that AMD will receive royalties on Sony's (NYSE:SNE) PS4 and Microsoft's (NASDAQ:MSFT) Xbox 1. I asked AMD for clarification, and they responded by stating, "Our target is 20% of revenue from semi-custom and embedded by Q4 ... but yes, both the Sony and Microsoft wins are for silicon sales and not licensing. Wii U is a licensing deal." So revenues are likely to significantly go up from previous generations of consoles. This is in line with the article posted on Barron's stating AMD would likely receive $60-$80 per APU. Based on data that points to flat PC sales this quarter, AMD's guidance of flat revenue, and console launches later this year, I suspect the next generation console wins will not show up until next quarter's financials.
The biggest risk I see for AMD would be another series of unexpected payments to Global Foundries. Given that PC sales are forecasted to pick up 2H 2013, and AMD entered into the third WSA amendment late last year (so the terms of the contract were negotiated while the PC market was declining) I do not believe a large payment is likely. Also, GloFo is ramping 28nm bulk silicon, which should help AMD meet their $1.15B target this year by producing Kaveri on GloFo 28nm, as reported by techpowerup.com. Techpowerup's article is backed by the Q&A session that followed the second WSA amendment call (specifically look at the line of questioning from Vivek Arya). Given this information, I am very interested in any news regarding the WSA that may come out Thursday.
The most likely risk for Q2, in my opinion, is lower revenue based on weak PC and GPU demand. Based on the reasons I have stated, if there is a revenue miss, I expect it will be small. If the numbers used from PC World and IDC are off, that would affect my analysis.
I will add that a miss at this point could be potentially worse than normal given AMD is already operating at a cash outflow. I still consider AMD a speculative play. If you believe in AMD's story, a miss on Thursday (depending on the tone of the call and guidance) could present a good buying opportunity. If you are long, I would not stray too far from a computer during the call.
During the call, other than guidance, I am most interested in news regarding the WSA with Global Foundries, Kaveri, a GPU refresh, workstation card sales, Kyoto sales, and anything regarding the HSA foundation.
Overall, my assessment for Thursday is essentially a repeat of the Q1 earnings call, but with non-GAAP and GAAP figures in a tighter spread due to the lack of the penalty associated with the Lonestar Campus sale/lease back. Yahoo has earnings at -$.13 a share, while earningwhispers.com has the estimate at $-.12, with the Whisper Estimate at -$.10. Given AMD is in a turnaround at this point, management meeting guidance is a win since it shows they are on track to return to positive cash flow in the back half of this year. The guidance set forth on Thursday, in my opinion, will be the focal point.