ADVANCED MICRO DEVICES AMD
October 19, 2011 - 6:57pm EST by
thoreau941
2011 2012
Price: 4.73 EPS $0.00 $0.75
Shares Out. (in M): 692 P/E 0.0x 6.5x
Market Cap (in $M): 3,300 P/FCF 0.0x 0.0x
Net Debt (in $M): 338 EBIT 0 0
TEV ($): 3,638 TEV/EBIT 0.0x 0.0x

Sign up for free guest access to view investment idea with a 45 days delay.

 

Description

What They Do:

  1. Computer Processors (~5.1 BN Revs)
    1. Low end computer processors
    2. High end server processors ("Bulldozer")
    3. Low voltage specialty processors for laptops ("Brazos") & ("Llano")
    4. Regular voltage specialty processors for desktops ("Llano" - higher voltage)
    5. Graphics processors (~1.7 BN Revs)
    6. Gaming, seismic simulation, multiple monitors
  2. License their IP (~50-100MM Revs)
    1. Graphics processors for Nintendo, Xbox etc.

Why Now:

  1. Street numbers and expectations have been destroyed by a negative preannouncement
  2. Analysts don't understand margin structure of AMD since divestment of large manufacturing segment - margins will be higher as they head towards being a design oriented company
  3. People who remember AMD before its structural change and think it is purely a back and forth between Intel and AMD for the most powerful chip
    1. The integration of graphics as part of the need in low power devices and higher density/utilization server platforms has changed the playing field in AMD's favor given their graphics tech - remember that AMD's graphics are industry leading, and while Intel is a generation ahead of them in processor tech.
    2. AMD is not trying to compete with Intel on the same playing field, simply put they have modularized their chip components so that they can easily adapt to the pressures that will come down the pike with ARM in the 2014/2015 timeframe - given the changing market dynamics, the underling design choices made by AMD will make them far more adaptable in the tech space over time.
  4. Earnings and expectation revisions likely to materialize in the first half of next year in an upward direction - folks brought down earnings materially with this quarter's preannouncement and have adjusted earnings down in the forward years commensurately and without consideration of increased ramps in chips etc.
  5. Expectations for the coming quarters are low, new product adoption is going to quickly drive results in several sequential quarters - long story short the new CEO is an ops guy, the product cycle is favorable for them, and there have been ample downgrades in the stock to allow for sequential beats - again doesn't change the valuation much but it gets folks interested in the stock again
  6. Intel results indicate that there is little to no semiconductor slowdown in consumer space contradictory to industry reports - for 3 straight quarters Intel has blown away the IDC numbers, AMD has as well (except for their bungled production in the last quarter). 

Quick Valuation:

1.       On a comps basis it is super cheap - not particularly relevant if you are looking at fundamentals but important to many...

2.       They are starting to generate some serious cash flow, in the forward years call it 400MM, so you are dealing in a 10% + yield

Balance Sheet/Leverage:

  1. Company has a topline of ~7BN
  2. Operating Profit of ~650MM
  3. Net debt of ~500MM
  4. FCF of ~400MM on a forward basis

Product Offerings:

1.       New server processors

a.       Bulldozer launch in Q3 ramping through Q4 and into the next update in Q212

b.      Their server share is ~6%...increasing it at all is good...each 1% of share gets them about ~0.07 cents in earnings

c.       New server product will be quickly adopted by existing customer base as it requires few incremental hardware upgrades (chip only can be slotted into existing boards)

d.      Cloud/shared resource computing customers are new and AMD product is better for their needs than existing intel products

2.       Consumer devices...

If you are/have losing/lost interest in this write-up by this point because of embedded biases for Intel, or cause you heard ARM chips are awesome and are going to dominate Intel and AMD, or if you just don't like tech because it is a low return on capital business with limited barriers to entry unless you have scale stick with me just a little longer....

Investing in the tech sector is never an easy task. Trying to show it as a value investment is an even harder one. AMD offers a very interesting opportunity to capture many of the major cycle changes and long-term developments that are going on in the technology space right now.

The first thing is the most important, prices for the processors or the "brains" of the computer are collapsing. (I will note at this point that Intel's gross margins are higher than AMD's do what you will with the info).  There are many reasons why prices are falling but some of the most important ones include the fact that computers are no longer to be an aggregation of component parts (a processor plus a graphics card, plus networking), but rather their headed towards becoming a system on one piece of silicon with the screen attached.

Demand for these devices is going to continue to grow and the billions of consumers (no exaggeration) that are currently priced out of the market are clamoring for low cost high processing power devices. Unlike in the US and other developed markets where were used to names such as Dell and HP often times the highest volume purveyors of technology in places like China and Brazil are local hardware builders. The moment lower price technology with similar capabilities is available those builders will gravitate towards that technology.

There are manufacturers such as ARM who produce designs for chips that have very simplistic processors in them. Simplistic is relative to the Intel's and AMD's of the world. The ARM chips don't have many of the legacy features which get built into chips over years and years of development. x86 designs (Intel and AMD bread and butter) have a million different legacy architectures stuffed in them from years of development and lots and lots of compatibility with tons and tons of pieces of hardware that are out there.

Hardware manufacturers want to use ARM hardware to make cheaper devices they can meet the needs of emerging market consumers. The difficulty is that much of the software that's out there does not run on ARM systems and many of the component hardware pieces that make a computer your computer don't work either (think audio drivers ethernet connectivity wireless chips etc.). Apple's iOS uses ARM in its closed software architecture, and Android ports to ARM, but remains light on productivity and connectivity applications.  The upcoming release of Microsoft Windows 8 (summer/fall 2012) will run on ARM hardware but there will be limited availability of software to run on those machines.

Long story short, the sweet spot for OEMs and system builders around the world would be if you could get a chip with all the x86 compatibility, the speed and horsepower of a high-end processor, very strong graphics capabilities to drive today's tablet and smartphone screens, all in a low-power package.  AMD has identified this as a key goal, Intel recently got on board by adjusting their product mix to address poor performance in their Atom chip.  AMD is leading the charge with its low power high throughput Zacate platform, and will soon be adding their Trinity platform to the mix.

A quick review of the platforms they have that fits this bill:

1.       Llano/Brazos is a new processor that integrates the processor part (math) with the graphics part (pictures/movies) - (Trinity will replace Llano in the 1st half of 2012, and yes, Llano is the chip on which they have had the low production yields)

a.       Putting them together in one package reduces costs to the OEM (HP, Dell, Lenovo, Acer, etc. want to sell more of their product)

b.      Saves battery life, has great marketing spin on graphics driving positioning to consumer

2.       The Llano is a "1st" in that the chips, while in one package act as discreet entities

a.       Intel's graphics capabilities are not discreet and use more of the less efficient processing (math processing) to get to less of a result

3.       This discrepancy will continue with the launch of the next generation Llano called "Trinity" and will continue to differentiate the platform even as Intel rolls out its next generation product in 1H12 know as "Ivy Bridge"

4.       OEM's NEED (not want) a new path to sell computers

a.       No longer market based on Ghz...market based on graphics and battery

b.      They get more margion in the products because they are able to cut out a large component cost and still get top tier performance

5.       AMD is a total laggard in the area until now...new shipments of Llano are not replacing old generations of AMD processors, they are taking share at higher margin

Software being developed from this point forward is beginning to take a massively different approach to the way it interfaces with hardware for two reasons.

1.       Instead of having one instruction the goes to the processor and is computed the software tools available to developers today are capable of automatically creating code that uses multiple parts of existing processors (multi-threaded processing takes advantage of multi core processors).

2.       In addition to the above software that commonly only uses the processor will now be able to have code executed across all the computers components because of the introduction of languages like C++ AMP, MSFT has built this and has given it away open source to developers and the manufacturers of developer's coding software). This type of computing is referred to as heterogeneous computing. The desktop PC you have today often has a processor and a graphics processor for example, when you go to click on a webpage in Internet Explorer 8 the webpage is rendered using the processor. If you click on the same webpage in Internet Explorer 9 the software is smart enough to use the processor, if the processor has multiple cores it uses more than one, and it allocates complicated floating-point calculations to your graphics processor which is far better suited to handle this type of math - faster, more efficient, and less power hungry software.

The implication of heterogeneous compute is that the software can now maximize the hardware resources that are available, your existing desktop is going to get more powerful without you doing anything to it or you can take much less powerful and much less power-hungry hardware and make a darn good computer out of it that can be smaller, handheld and much lower cost for people who want that type of product.

AMD has a highly competitive offering that takes advantages of these changes, due to its integrated graphics capabilities and low power offerings.

Licensing revenues:

1.       AMD has a royalty deal with Nintendo for the new game console that will be built out in 2012, Nintendo will pay AMD for graphics IP, the original Wii was worth ~500MM to the company over 4 years, the analyst numbers contemplate only about 15% of that; numbers will likely be adjusted accordingly.

a.       There are ~75MM installed Wiis in the world, and growing (at a slow rate) the next generation consoles will be HD, needing more robust graphics

2.       The company will likely win bids for any next generation video processing in gaming consoles, Nvdia (graphics competitor has focused on compute products away from graphics); the PS3 revenues were over 1BN USD for Nvidia when it was release in the earlier part of the decade.

a.       Clearly the market for gaming has evolved towards ipads and ipods etc, but the need for high end graphics in consoles is extremely important as Microsoft (xbox) and Sony (Play Station) have an installed base of ~85MM consoles, and the next generations will have an expected die of 5-7 years and the need to unify entertainment in HD (think about the MSFT deal with Fios/Comcast to replace cable boxes etc), gaming and other processing - The AMD platforms are the clear winning option for this kind of roll out.

IP etc.

1.       Quick and dirty, AMD has a massive amount of IP for everything from processors (1.5BN settlement last year with Intel), to memory controllers, to next generation graphics.

2.       They have comfortable relationships with AAPL and Intel at this point, but have in recent years split their products out on a development level, their cores, memory controllers, GPUs etc. have all had their interface languages put out in the open for folks to use...and to adopt for eventual use in new products.

3.       The company is transitioning to having a large part of its revs in the new mobile world come from IP just like ARM.

4.       Once they throw together a couple of quarters of decent earnings, and the momentum picks up in the stock, many of the sell side will introduce this concept to pump up valuation - right now nobody cares. 

A couple of outcomes:

1.       In a base case scenario:

a.       With decent volumes and higher margins associated with the new products and server share, 2012 EPS will likely be ~.75 (vs. 0.59 est.)

2.       Upside Case

a.       Upside case will get 2012 eps closer to ~1.00 based on higher shipments, the new introductions of GPUs

b.      Addition of royalty revenue in 2012 from new Nintendo build out can be an additional .15 - .25 cents that is not integrated into analyst numbers

How you can get creamed owning AMD:

1.       The company doesn't control its foundry and delays in 3rd party manufacturing could have impact on the ability to ship chips;

a.       Well, that already happened for their 32nm offerings

2.       Consumers are not interested in the newer release of the trinity processor;

a.       unlikely given the demand for more graphic intensive applications (Flash, Youtube, HTML5 - drives web content using graphics and needs commensurate graphics processor power)

3.       Intel could take action to squash AMD margins by lowering their prices, they simply have greater resources;

a.       unlikely as AMD and Intel both play against new ARM based designs and fundamental price cuts on the x86 side (current product) would be painful for Intel given new competitor on the roadmap for 2014

4.       Capital is deployed in an equity which is subject to larger semi cycle which may be slowing down;

Catalyst

Q4 margins and shipments higher than expected
Buyout on IP portfolio
Reassurance that low shipment volumes are not permanent on Q3 call
 
 
    sort by   Expand   New

    Description

    What They Do:

    1. Computer Processors (~5.1 BN Revs)
      1. Low end computer processors
      2. High end server processors ("Bulldozer")
      3. Low voltage specialty processors for laptops ("Brazos") & ("Llano")
      4. Regular voltage specialty processors for desktops ("Llano" - higher voltage)
      5. Graphics processors (~1.7 BN Revs)
      6. Gaming, seismic simulation, multiple monitors
    2. License their IP (~50-100MM Revs)
      1. Graphics processors for Nintendo, Xbox etc.

    Why Now:

    1. Street numbers and expectations have been destroyed by a negative preannouncement
    2. Analysts don't understand margin structure of AMD since divestment of large manufacturing segment - margins will be higher as they head towards being a design oriented company
    3. People who remember AMD before its structural change and think it is purely a back and forth between Intel and AMD for the most powerful chip
      1. The integration of graphics as part of the need in low power devices and higher density/utilization server platforms has changed the playing field in AMD's favor given their graphics tech - remember that AMD's graphics are industry leading, and while Intel is a generation ahead of them in processor tech.
      2. AMD is not trying to compete with Intel on the same playing field, simply put they have modularized their chip components so that they can easily adapt to the pressures that will come down the pike with ARM in the 2014/2015 timeframe - given the changing market dynamics, the underling design choices made by AMD will make them far more adaptable in the tech space over time.
    4. Earnings and expectation revisions likely to materialize in the first half of next year in an upward direction - folks brought down earnings materially with this quarter's preannouncement and have adjusted earnings down in the forward years commensurately and without consideration of increased ramps in chips etc.
    5. Expectations for the coming quarters are low, new product adoption is going to quickly drive results in several sequential quarters - long story short the new CEO is an ops guy, the product cycle is favorable for them, and there have been ample downgrades in the stock to allow for sequential beats - again doesn't change the valuation much but it gets folks interested in the stock again
    6. Intel results indicate that there is little to no semiconductor slowdown in consumer space contradictory to industry reports - for 3 straight quarters Intel has blown away the IDC numbers, AMD has as well (except for their bungled production in the last quarter). 

    Quick Valuation:

    1.       On a comps basis it is super cheap - not particularly relevant if you are looking at fundamentals but important to many...

    2.       They are starting to generate some serious cash flow, in the forward years call it 400MM, so you are dealing in a 10% + yield

    Balance Sheet/Leverage:

    1. Company has a topline of ~7BN
    2. Operating Profit of ~650MM
    3. Net debt of ~500MM
    4. FCF of ~400MM on a forward basis

    Product Offerings:

    1.       New server processors

    a.       Bulldozer launch in Q3 ramping through Q4 and into the next update in Q212

    b.      Their server share is ~6%...increasing it at all is good...each 1% of share gets them about ~0.07 cents in earnings

    c.       New server product will be quickly adopted by existing customer base as it requires few incremental hardware upgrades (chip only can be slotted into existing boards)

    d.      Cloud/shared resource computing customers are new and AMD product is better for their needs than existing intel products

    2.       Consumer devices...

    If you are/have losing/lost interest in this write-up by this point because of embedded biases for Intel, or cause you heard ARM chips are awesome and are going to dominate Intel and AMD, or if you just don't like tech because it is a low return on capital business with limited barriers to entry unless you have scale stick with me just a little longer....

    Investing in the tech sector is never an easy task. Trying to show it as a value investment is an even harder one. AMD offers a very interesting opportunity to capture many of the major cycle changes and long-term developments that are going on in the technology space right now.

    The first thing is the most important, prices for the processors or the "brains" of the computer are collapsing. (I will note at this point that Intel's gross margins are higher than AMD's do what you will with the info).  There are many reasons why prices are falling but some of the most important ones include the fact that computers are no longer to be an aggregation of component parts (a processor plus a graphics card, plus networking), but rather their headed towards becoming a system on one piece of silicon with the screen attached.

    Demand for these devices is going to continue to grow and the billions of consumers (no exaggeration) that are currently priced out of the market are clamoring for low cost high processing power devices. Unlike in the US and other developed markets where were used to names such as Dell and HP often times the highest volume purveyors of technology in places like China and Brazil are local hardware builders. The moment lower price technology with similar capabilities is available those builders will gravitate towards that technology.

    There are manufacturers such as ARM who produce designs for chips that have very simplistic processors in them. Simplistic is relative to the Intel's and AMD's of the world. The ARM chips don't have many of the legacy features which get built into chips over years and years of development. x86 designs (Intel and AMD bread and butter) have a million different legacy architectures stuffed in them from years of development and lots and lots of compatibility with tons and tons of pieces of hardware that are out there.

    Hardware manufacturers want to use ARM hardware to make cheaper devices they can meet the needs of emerging market consumers. The difficulty is that much of the software that's out there does not run on ARM systems and many of the component hardware pieces that make a computer your computer don't work either (think audio drivers ethernet connectivity wireless chips etc.). Apple's iOS uses ARM in its closed software architecture, and Android ports to ARM, but remains light on productivity and connectivity applications.  The upcoming release of Microsoft Windows 8 (summer/fall 2012) will run on ARM hardware but there will be limited availability of software to run on those machines.

    Long story short, the sweet spot for OEMs and system builders around the world would be if you could get a chip with all the x86 compatibility, the speed and horsepower of a high-end processor, very strong graphics capabilities to drive today's tablet and smartphone screens, all in a low-power package.  AMD has identified this as a key goal, Intel recently got on board by adjusting their product mix to address poor performance in their Atom chip.  AMD is leading the charge with its low power high throughput Zacate platform, and will soon be adding their Trinity platform to the mix.

    A quick review of the platforms they have that fits this bill:

    1.       Llano/Brazos is a new processor that integrates the processor part (math) with the graphics part (pictures/movies) - (Trinity will replace Llano in the 1st half of 2012, and yes, Llano is the chip on which they have had the low production yields)

    a.       Putting them together in one package reduces costs to the OEM (HP, Dell, Lenovo, Acer, etc. want to sell more of their product)

    b.      Saves battery life, has great marketing spin on graphics driving positioning to consumer

    2.       The Llano is a "1st" in that the chips, while in one package act as discreet entities

    a.       Intel's graphics capabilities are not discreet and use more of the less efficient processing (math processing) to get to less of a result

    3.       This discrepancy will continue with the launch of the next generation Llano called "Trinity" and will continue to differentiate the platform even as Intel rolls out its next generation product in 1H12 know as "Ivy Bridge"

    4.       OEM's NEED (not want) a new path to sell computers

    a.       No longer market based on Ghz...market based on graphics and battery

    b.      They get more margion in the products because they are able to cut out a large component cost and still get top tier performance

    5.       AMD is a total laggard in the area until now...new shipments of Llano are not replacing old generations of AMD processors, they are taking share at higher margin

    Software being developed from this point forward is beginning to take a massively different approach to the way it interfaces with hardware for two reasons.

    1.       Instead of having one instruction the goes to the processor and is computed the software tools available to developers today are capable of automatically creating code that uses multiple parts of existing processors (multi-threaded processing takes advantage of multi core processors).

    2.       In addition to the above software that commonly only uses the processor will now be able to have code executed across all the computers components because of the introduction of languages like C++ AMP, MSFT has built this and has given it away open source to developers and the manufacturers of developer's coding software). This type of computing is referred to as heterogeneous computing. The desktop PC you have today often has a processor and a graphics processor for example, when you go to click on a webpage in Internet Explorer 8 the webpage is rendered using the processor. If you click on the same webpage in Internet Explorer 9 the software is smart enough to use the processor, if the processor has multiple cores it uses more than one, and it allocates complicated floating-point calculations to your graphics processor which is far better suited to handle this type of math - faster, more efficient, and less power hungry software.

    The implication of heterogeneous compute is that the software can now maximize the hardware resources that are available, your existing desktop is going to get more powerful without you doing anything to it or you can take much less powerful and much less power-hungry hardware and make a darn good computer out of it that can be smaller, handheld and much lower cost for people who want that type of product.

    AMD has a highly competitive offering that takes advantages of these changes, due to its integrated graphics capabilities and low power offerings.

    Licensing revenues:

    1.       AMD has a royalty deal with Nintendo for the new game console that will be built out in 2012, Nintendo will pay AMD for graphics IP, the original Wii was worth ~500MM to the company over 4 years, the analyst numbers contemplate only about 15% of that; numbers will likely be adjusted accordingly.

    a.       There are ~75MM installed Wiis in the world, and growing (at a slow rate) the next generation consoles will be HD, needing more robust graphics

    2.       The company will likely win bids for any next generation video processing in gaming consoles, Nvdia (graphics competitor has focused on compute products away from graphics); the PS3 revenues were over 1BN USD for Nvidia when it was release in the earlier part of the decade.

    a.       Clearly the market for gaming has evolved towards ipads and ipods etc, but the need for high end graphics in consoles is extremely important as Microsoft (xbox) and Sony (Play Station) have an installed base of ~85MM consoles, and the next generations will have an expected die of 5-7 years and the need to unify entertainment in HD (think about the MSFT deal with Fios/Comcast to replace cable boxes etc), gaming and other processing - The AMD platforms are the clear winning option for this kind of roll out.

    IP etc.

    1.       Quick and dirty, AMD has a massive amount of IP for everything from processors (1.5BN settlement last year with Intel), to memory controllers, to next generation graphics.

    2.       They have comfortable relationships with AAPL and Intel at this point, but have in recent years split their products out on a development level, their cores, memory controllers, GPUs etc. have all had their interface languages put out in the open for folks to use...and to adopt for eventual use in new products.

    3.       The company is transitioning to having a large part of its revs in the new mobile world come from IP just like ARM.

    4.       Once they throw together a couple of quarters of decent earnings, and the momentum picks up in the stock, many of the sell side will introduce this concept to pump up valuation - right now nobody cares. 

    A couple of outcomes:

    1.       In a base case scenario:

    a.       With decent volumes and higher margins associated with the new products and server share, 2012 EPS will likely be ~.75 (vs. 0.59 est.)

    2.       Upside Case

    a.       Upside case will get 2012 eps closer to ~1.00 based on higher shipments, the new introductions of GPUs

    b.      Addition of royalty revenue in 2012 from new Nintendo build out can be an additional .15 - .25 cents that is not integrated into analyst numbers

    How you can get creamed owning AMD:

    1.       The company doesn't control its foundry and delays in 3rd party manufacturing could have impact on the ability to ship chips;

    a.       Well, that already happened for their 32nm offerings

    2.       Consumers are not interested in the newer release of the trinity processor;

    a.       unlikely given the demand for more graphic intensive applications (Flash, Youtube, HTML5 - drives web content using graphics and needs commensurate graphics processor power)

    3.       Intel could take action to squash AMD margins by lowering their prices, they simply have greater resources;

    a.       unlikely as AMD and Intel both play against new ARM based designs and fundamental price cuts on the x86 side (current product) would be painful for Intel given new competitor on the roadmap for 2014

    4.       Capital is deployed in an equity which is subject to larger semi cycle which may be slowing down;

    Catalyst

    Q4 margins and shipments higher than expected
    Buyout on IP portfolio
    Reassurance that low shipment volumes are not permanent on Q3 call
     
     

    Messages


    SubjectRE: Negative FCF?
    Entry10/21/2011 10:51 AM
    Memberthoreau941

    cuyler, thanks for the comment.

    It seems Ycombinator has already pointed out the financing agreement between IBM and AMD that has made their FCF look a little funny when you use purely quantitative screens etc. Please reference his/her reconciliation in the comment board. What is really nice is that many investors and quant screens will start to see the FCF generation in their "back of the envelope" calculations over the next year. I would also note that the company has done a great job building cash on the balance sheet, and as a result, has materially reduced its net leverage over the last year. Please throw up any specific questions about the FCF reconciliation if you have them and I will be glad to help out if I can.

    I understand your question about why the company is a good one to own - I think it is the right one to ask. I am making a little bit of an assumption that you are asking the question in a separate context from your FCF question, so forgive me if I have assumed incorrectly.

    I am happy to own the company because I view it to have some really nice skew to the upside BOTH operationally and as it relates to market perception. It is so beaten down that small incremental changes in their performance at the factory and in the product reception by OEMs will drive great value in terms of earnings etc. (small clicks in server share lead to big relative chunks of EPS, incremental design wins for higher margin llano/trinity end up leveraging their marketing spend etc.) The nice thing is that the new management team seems to understand that there is opportunity to deliver in a way that has come to be unexpected by BOTH customers and Mr. Market. In terms of how things can go incrementally wrong, I mentioned a bunch of the risks in the write-up, and make no mistake, they are real, but the valuation takes into account what is effectively a terminal lifecycle for AMD over the next 2-3 years. Given that their entire product roadmap is dedicated to designing components that can eventually be peeled off and included in the SOC designs that are sure to dominate the market in the future I believe predicting their demise based on what is currently running on the assembly line is foolish (there are other x86 manufacturers for which this extrapolation may be more appropriate without huge incremental R&D and acquisitions). So long story short, operational improvements will translate to numbers, and it is generally an unexpected positive change where it doesn't take much to move the needle.

      Back to top