超微半導體 (AMD) 2009 Q3 法說會逐字稿

Good afternoon.

I'll be your conference operator for today.

At this time, I would like to welcome everyone to AMD's third quarter 2009 earnings conference call.

All lines have been placed on a listen-only mode at this time.

After the speakers' remarks you will be invited to participate in the question-and-answer session.

As a reminder, this conference is being recorded today.

I would now like to turn the conference over to Ms.

Ruth Cotter, director of investor relations for AMD.

Please go ahead.

Thank you and welcome to AMD's third-quarter earnings conference call.

Our participants today are Dirk Meyer, our President and CEO, and Bob Rivet, our Chief Operations and Administrative Officer.

This is a live call and will be replayed via webcast on AMD.com.

There will also be a telephone replay.

The replay number is 888-258-7854.

Outside of the United States the number is 703-925-2490.

The access code for both is 1397389.

The telephone replay will be available for the next ten days starting later this evening.

I would like to take this opportunity to remind everyone that AMD will hold its financial analyst day on November 11th at its headquarters in Sunnyvale, California.

I would also like to bring to your attention that our Q4 2009 earnings quiet time will begin at the close of business on Friday, December 11th.

References to AMD on this earnings call refer to AMD Product Company, which refers to the operating results of AMD excluding our foundry segment and inter segment elimination.

We provide a reconciliation of AMD Product Company to AMD consolidated operating results, which are reported for GAAP purposes.

in today's earnings press release issued earlier this afternoon.

Before we begin today's call, I would like to caution everyone that we will be making forward-looking statements about managements expectations.

Investors are cautioned those statements are based on current beliefs, assumptions and expectations, speak only as of the current date and involve risks and uncertainties that could cause actual results to differ materially from our current expectations.

The semiconductor industry is generally volatile and market conditions are particularly difficult to forecast, especially in light of the current state of the economy.

We encourage you to review our filings with the SEC where we review the risk factors setting forth information that could cause actual results to differ materially from our expectations.

You will find detailed discussions about such risk factors in our most-recent SEC filings, AMD's quarterly earnings report on Form 10-Q for the quarter-ended June 27, 2009.

With that, I'd like to turn the call over to Dirk.

Dirk?

Thanks, Ruth, and thanks to everybody on the phone for joining us today.

Before I begin, I'd like to introduce the newest member of our management team, Mr.

Thomas Seifert, who is taking over the CFO role, Thomas, welcome aboard.

Thank you, Dirk, I'm excited to be here and I look forward to getting to work with all of you out there in the years ahead beginning with our upcoming financial analyst day in November.

And -- so again, welcome Thomas and I also want to thank Bob for your steady hand in this arena over the past nine years.

I know you're really looking forward to having 100% focus on your new role as Chief Operations and Administrative Officer.

Turning now to the business.

Third quarter consumer PC demand continued to improve from prior periods, with particular strength in notebooks and in China and continued recovery in Europe and North America, and it appears the commercial IT markets are positioned to improve next year.

We experienced very strong global demand for our discrete graphics offerings, particularly in the notebook sector, as consumers continue to value a graphics-rich PC experience, and I'm really encouraged by our financial performance in the quarter.

We grew revenues by 18% sequentially, w improved our gross margins through a continued transition to 45nm CPUs, increased factory utilization and higher CPU ASPs, and we controlled expenses very well, reducing OpEx by 9%, and most importantly, AMD, The Product Company put a positive number on the bottom line.

Meanwhile, we delivered on each of the three key pillars of our differentiation; platforms, discrete graphics leadership, and marketing focused on real-world end-user needs.

With respect to platforms, our momentum continues to build with HP, Acer, Toshiba, Sustech and MSI all announcing plans to introduce more than 70 notebooks based on AMD's latest generation of mainstream and ultra-thin platforms.

In servers, HP, IBM and Dell are all using our new six-core AMD Opteron processors to meet the growing demand for energy-efficient and high-density server platforms.

We now offer 15 full-featured low and very low power options for cloud computing deployments and hyper scale data centers.

Our graphics business continues to gain momentum as we added to our industry-leading discrete graphics portfolio.

The Radeon HD 5870 captured the graphics performance title, winning greater than 50 industry awards up to this point.

The Radeon HD 5000 family is the first in the industry designed to take advantage of Microsoft's next generation Direct X11 technology.

We believe this new offering is uniquely well positioned to take advantage of Microsoft's Windows 7 upcoming rollout, and the HD 5000 family also supports our widely-acclaimed Eyefinite multi-display technology allowing a single PC to drive up to six high def monitors.

In marketing we launched a pair of new important programs in the quarter; VISION technology from AMD and the AMD Fusion Partner Program.

The Fusion Partner Program will help AMD's channel partners gain sales traction and streamline the delivery of AMD platforms into the channel and with our new VISION campaign we introduced a new approach to retail merchandising, simplifying the consumer buying experience by highlighting what can be done with a PC, rather than simply what's inside the PC.

Industry response to the program has been very positive indeed.

Both of these new programs are designed to demonstrate and reinforce the unique value proposition of an all AMD platform.

On the manufacturing front we joined with Globalfoundries to break ground on Fab 2 in New York, a state-of-the-art semiconductor manufacturing facility that will provide additional leading-edge manufacturing capacity when the facility enters production, which is scheduled for 2012.

In summary the AMD value proposition continues to resonate across segments, product lines and geographies.

Our new products are strong and in the case of our graphics offering, absolutely stunning.

Our marketing programs are focused on the things that demonstrate our differentiation; graphics leadership, balanced platforms, and performance per watt superiority.

Our customers tell me they like what we're delivering today, as well as what they see coming in the future.

Going forward we believe we are well positioned to succeed.

We are focused on the biggest and most opportunity-rich market for our technology, the PC market.

Our product momentum is increasing, we are winning confidence and enthusiasm from our world-class customers, and we will continue to carefully manage costs, increasing operating efficiencies, and improve our financial performance.

With that, over to you, Bob.

Thank you, Dirk, and welcome, Thomas.

In the fourth quarter of last year we outlined a breakeven business model for AMD, The Product Company that was both sized to the new market reality and optimized for agility and flexibility and in the past quarter we demonstrated we are there, on the top line and on our cost structure, achieving profitability in each of our core businesses.

Meanwhile, our partner, Atec, reinforced their strategic committment to the Globalfoundries investment opportunity by choosing to fully fund their first capital call, which took place during the past quarter.

In short, we hit all of the major milestones in the quarter.

Third party -- third quarter revenue was $1.396 billion, growing 18% compared to the second quarter of 2009.

Compared to the third quarter of last year revenue declined 22%; however, excluding the $191 million process technology license revenue, revenue declined 13% on a non-GAAP basis.

AMD Product Company reported a non-GAAP net income of $2 million in the third quarter of 2009, which excludes the net favorable impact of $54 million primarily from a $66 million gain from the repurchase of debt.

Third quarter non-GAAP operating income was $47 million, which excludes the impact of unusual items as described in the press release.

We made very good progress on gross margin in the third quarter, a result of higher unit volumes across the entire portfolio, a stronger mix of 45nm CPUs, higher micro processor ASPs, and improved factory utilization rates.

As a result, third quarter non-GAAP gross margin improved to 38% compared to 27% in the prior quarter.

You may remember our breakeven model is predicated on SG&A of $200 million and R &D of $300 million per quarter.

We met and exceeded both targets in the third quarter reducing SG&A spending to $195 million and R&D to $285 million.

As a result of our operating discipline we saw our adjusted EBITDA improve significantly over the second quarter to $214 million in the third quarter.

Switching to the business segments, we were pleased to report that both of our Computing Solutions and Graphic business were profitable in the third quarter, each reporting sequential double-digit revenue growth.

We were encouraged by our demand for our products and the growth of our business with OEM and channel partners, particularly in Europe and greater China.

Computing Solution revenue was $1.069 billion in the third quarter, up 17% sequentially, driven by higher micro processor ASPs and double-digit unit growth.

Unit growth was mainly driven by notebook processor shipments, which increased by 28% quarter over quarter.

We saw revenue and unit growth in each of our desktop, notebook and server businesses and saw a record quarter in our chipset business.

Operating income for the Computing Solution group was $76 million, a significant increase over a loss of $72 million in the second quarter.

In the Graphics segment revenue for the quarter was $306 million, up 22% sequentially.

Growth was driven by significantly higher shipments of mobile discrete graphic processors, demand in the add-in board and retail channels was spurred by the launch of the Radeon HD 5000 series, the industry's first DX11 compatible graphics cards, and the Graphics business also reported an operating income of $8 million in the quarter, an improvement from a loss of $12 million from the second quarter.

Now turning to the AMD Product Company's balance sheet, our cash and marketable security balance at the end of the quarter was $1.5 billion.

We used $104 million to repurchase $186 million of principal amount of debt.

Excluding this retirement of debt cash was flat quarter on quarter.

Since this time last year we have retired a total of $419 million of outstanding convertible notes.

As I mentioned a moment ago, it's worth noting that Globalfoundries had its first capital call in July for $260 million.

AMD declined to participate in this funding.

Now let me turn to the outlook.

The following statements concerning AMD, The Product Company are forward-looking and the actual results could differ materially from current expectations.

For the fourth quarter AMD expects revenue to be up modestly.

We expect operating expenses to be approximately $500 million, in line with the breakeven model, and capital expenditure guidance remains unchanged at approximately $100 million for the year.

In conclusion, we executed well in the third quarter, reaching our breakeven model and achieving profitability while we continue to improve our balance sheet.

We believe we are now operating at a grounded and sustainable business model, the benefits of which will be amplified with a return to normal IT spending in the commercial sector.

In the meantime, demand for our balanced platforms, discrete graphics and chipsets offering should provide continued strong momentum as we go throughout the year.

At this point, let's go to Q&A with Ruth.

Thanks, Bob.

We'll now move to our question-and-answer session, which Dirk and Bob will host.

Huey, if you could please poll our participants for questions that would be good.

Thank you.

Thank you, ma'am.

(Operator Instructions).

Our first question comes from Doug Freedman from Broadpoint.

Your question, please?

Great, thanks for taking my question.

If I could start off with I think the language we understand as seasonality, if you could help us understand where moderately fits in with seasonality on the top-line guidance?

Sure, Doug.

Our view is that seasonal increases Q3 to Q4 average around 9% and are 6% or 7% on the low side.

Given the big build we've seen of PCs in anticipation of the Win 7 launch we're really wanting to see the consumption side of that build happen in Q4 and therefore at this point we're calling moderate, which is a little bit less bullish than average seasonality I would say.

All right, terrific, thanks for that color.

Can you talk a little bit about what are going to be the give and takes on the gross margin and if you wouldn't mind in the process of doing that maybe rank for us the gross margin contribution differences between GPU notebook and desktop platforms, maybe lastly fit server in there, what our expectation and outlook should be on the gross margin line.

How are you going to balance gross margins with possible share gains or potential opportunities for share gains?

Sure.

First clarifying point, your question about the product,-by-product break down is directional meaning Q2 to Q3 over the next quarter?

I'm looking -- more helpful Q3 going to Q4.

Okay.

Well standing back, first of all the Q2 to Q3 improvement was driven by the three factors that we've been talking about and talked about on the last call; number one was increase in mix of 45nm CPUs, number two factory utilization, number three ASPs.

We certainly got the opportunity to drive forward-looking improvement based on the first of those going from Q3 to Q4.

We're not done completely on the 45nm transition with a sales-out business -- sales-out basis.

As well we'll drive slightly higher utilization rates in Q4 as compared to Q3.

ASPs are of course always a wild card but we feel good about some of the new products that we introduced, both on the GPU side, as well the CPU side with the new notebook programs that are hitting the market.

You asked also about how we think about gross margin playing off against market share, and there we're going to approach the market the way we always have, which is aimed at maximizing gross margin dollar accumulation over the course of the quarter.

And this is Bob, and I just would amplify, and particularly since we still have a lot of capacity in our factories to take advantage of upside unit opportunities obviously we'll do the appropriate math on each of those opportunities, as Dirk said, but we have definitely room to ship more units.

Can you offer us some sort of a timeline as well on separation and deconsolidation between AMD, The Product Company and Globalfoundries and what milestones we might measure along that path?

Yes, if we circle back, we've had this conversation before so I'll remind you, it's really this simple.

The biggest milestone will be third-party revenue by Globalfoundries, but it will -- from that perspective and they have a ways to go before that'll happen but it's really the appropriate magnitude of third-party revenue so it's not a related-party transaction, just between Globalfoundries and AMD.

There are other accounting things that will have to happen but that's the single biggest thing that I would say would be the fencepost that needs to be out there and measured over time.

Terrific and congratulations on the profitable quarter.

Thank you.

Thank you, sir.

Our next question comes from Ross Seymore with Deutsche Bank.

Your question, please?

Hi guys.

Just wanted to get a little more color on the ASP trends that you highlighted in the quarter, both within CPUs and GPUs, just what drove those between the various metrics, mix, apples-to-apples pricing, et cetera?

Sure.

First on the CPU side, the biggest driver was in the desktop part of the product line, largely driven by mix improvement, relatively larger mix of both Phenom 2 and Athlon 2 products out there.

Notebooks largely flat.

Server ASPs were actually up a bit in the quarter,although the overall server revenue grew more or less in line with MPU revenue.

On the GPU side, as we said, I think, in the press release, ASPs were down just a bit quarter on quarter.

And I would have thought with some of the new products that you had on the GPU side that that would have helped lend up ASPs.

Is it just they didn't have a chance to get -- to be a meaningful part of the mix to impact overall ASPs?

Yes, exactly right.

The 5000 series, in particular, really didn't contribute meaningfully in the Q3 timeframe.

So if you think about it going forward -- I know it's always hard to talk about ASPs in the future -- but just from the mix component in GPUs would you expect that to have a positive impact as you go into the fourth quarter?

It's certainly an opportunity, yes.

And a follow-on question, completely different topic.

Just from an SOI technology point of view and with your new marketing program with Fusion, et cetera, how should we think about going from SOI to both CMOS from a timing perspective, if at all?

Yes, it's a good question and we'll put a little bit more light on this in the analyst call.

Of course the CPU products are on 45nm SOI today.

We'll have CPU so-called Fusion parts in 32nm SOI in the next generation and then the bulk SOI question is one we evaluate in every subsequent generation.

More to come in the analyst conference.

Okay, great.

Thank you.

Thank you, sir.

Our next question comes from Chris Danely with JPMorgan.

Your question, please.

Thanks, guys.

Real quick, Bob, just to follow up on Doug's question, can you give us a revenue number or level for the Globalfoundries business and does that include the charter revenue or is it just Globalfoundry revenue for when you can start to spin it out?

It's just Globalfoundry revenue, and that's a little bit of.

I'll call it.

the mystery of how big that number has to be, but it has to be a respectable portion of what Globalfoundry would report, so I'd say it's got to probably be above 10%, 15% and it would be helpful if it's more than just one additional customer.

So it's got to be a meaningful number.

It can't just be $1 million, it's got to be a bigger number than that.

Sure.

And then Dirk, any plans for a netbook processor offering?

Yes,netbook is a little bit of a nebulous term.

We actually have customers who have, I'll call it netbook-like offerings based on our components out there today, meaning screen sizes down below 12 inches and we continue to work with our customers to enable, I'll say, small, thin and light form factors.

You'll hear more in the financial analyst conference next month about the product line up that we'll be rolling out over the course of the next two years, which will include increasing focus on those small form factor notebooks.

Okay, great.

And my last question, you mentioned a little bit of a build for Win 7 out there, kind of goes against what some of your competitors are saying, can you just shed a little more color on that and maybe give some clarity as to the magnitude and where the product is?

Yes, that's a good question.

I didn't mean to convey necessarily inventory builds in the channels, rather, simply a lot of PCs have been built.

We're not hearing at this point about any unnatural pile up of inventory, but nonetheless there's a lot of consumption that still has to happen in Q4 before we can say we're really entering a normal seasonal sell-in cycle on Q4 in preparation for first-half next year PC consumption.

Great.

Thanks a lot.

Sure.

Thank you, sir.

Our next question comes from Glen Yeung with Citi.

Please go ahead.

Thanks.

Can you give us a sense looking in the fourth quarter whether or not you think graphics or your microprocessor business will grow faster?

If you again go back, let's just talk about seasonality in general and then get into specifics but in typical seasonality mode micro processors lead.

They have the strongest growth rate from third to fourth, the big quarter for graphics is the third.

Now as Dirk outlined, we're not sure we're exactly normal yet.

I would say that tren -- so therefore -- but we don't think that'll break the trend in the fourth quarter.

We expect CPU revenues to be stronger than GPU between the two pieces.

Okay, makes sense.

And then you mentioned that you're starting to see Europe get a little bit better here and that's actually been a region that you've done relatively well in.

Do you think then as Europe comes back given the product portfolio that you have that you can continue to show good growth as you maintain or even gain share in that market?

Yes, I would say in Europe is fairly an important market for us.

One because we have had relatively good share in that region over the years and number two, it tends to be a relatively high attach-rate region for discrete GPUs, which is also important for us.

So as Europe comes back that's going to help us differentially and particular Eastern Europe happens to be a very rich discrete graphics market.

Okay, and last for me, Dirk, maybe for you, if you could just give us some examples of where you're gaining traction in -- with Congo and with Istanbul, what kind of business we might expect to see?

And then maybe add to that what the OEMs are telling you about those products.

Are they coming to you because they're great products, or are they coming to you because they feel like you need to gain more share here?

What's driving them towards your product at this point?

Good question, and I would say both.

Clearly, all of our customers want to -- strong suppliers in their supply base and that's a positive factor for us but as well our products are darn good products.

In particular, Istanbul is doing well in an area that's been one of historic strength for us, mainly four and eight-socket servers.

The other area where we're seeing a lot of opportunity along side our OEMs is in high-density data center installations where our EE and HE Opteron parts in the six core variety are really darn good.

On the Congo, or second generation thin and light platforms, we've got a number of design wins, as I spoke to in the opening and remarks.

I expect to see a lot more momentum in that essentially new product category that we created between netbooks and mainstream notebooks going into the Christmas cycle.

We created that category really in partnership with HP.

We only had one platform in the market together was HP DV2, it did well relative to our expectations, but we'll have a much broader assortment of platforms walking into the Christmas cycle and I expect we'll see increased momentum as a result.

That's great.

Good job, you guys.

Thank you.

Thank you, sir.

Our next question comes from David Wong with Wells Fargo.

Thank you very much.

Can you give us some idea of what Istanbul was a percentage of server chips in the third quarter and what you expect for the fourth quarter, and did you see growth -- was growth higher in four-way server processes than two-way server processors in September?

Sure.

High level the six-core stuff was about a third of our server mix overall and I would call the growth of our two and four-socket platforms to be first order similar.

Thanks very much.

Thank you, sir.

Our next question comes from JoAnne Feeney with FTN Equity Capital.

Please go ahead.

Hi, good afternoon, folks, and congrats on a very nice quarter.

Thank you.

I'd like some more clarity, if you would, on the notebook side.

The strength there is a little bit of a surprise given that you aren't launching Tigress until this quarter so if you could clarify that was -- that did not include any sales with Tigress, correct, and if you could explain where that strength came from I think that 'd be a great help.

Sure, JoAnne.

Actually we did start selling Tigress components into OEMs at the very end of the quarter to prepare them for their builds for the Christmas cycle, but to your point most of the sales was on the prior-generation Puma platform where we just saw a very strong back-to-school cycle., so as a result there were good notebook sales in the quarter.

Well, that's terrific news.

And then if I could a follow up on the graphic side.

What do you see as we transition over to your newer product, the 5000 series, what do you see as your gross margin potential in that segment and how does that evolve over the next several quarters as you move from OEM sales to add-in board sales?

Think about it this way, JoAnne, is in the graphics business, the zone, the zip code you want to be in is in the 35% to 45% gross margin level.

We're not in that model right now.

We believe the execution we've had in the 5000 series will help us push toward closer to the model, at least break in from the lower level, so it's moving in the appropriate direction.

Do you feel like prices are set for these new products where you can get there once those take over your offerings, or is it another generation that we should expect to wait before that really happens?

It's hard to be that clairvoyant, JoAnne.

I'll give you a few factors to watch.

One is we've been doing extremely well on the current series with our OEM customers in notebook and desktop.

The big opportunity is for us to further increase our share in the so-called AIB channel, and in particular drive the high-end GPU business in that channel.

We feel really bullish about the 5000 series as we walk into Q4 and obviously next year.

The next big opportunity for us is in workstation graphics, which we've spoken to in the past.

There the rate of penetration is going to be naturally slower since that's an enterprise market and it takes us time to get the ISV certifications, ISV-(inaudible) marketing programs going and so on, and I think you're looking at a couple of years story to really get all of the way into the zone that Bob described.

Okay, and then just one follow up that -- on someone else's question about the budget thin & light.

Do you have a sense of what share of your fourth quarter notebook shipments that might be able to achieve for the Congo platform?

Sure, well less than half.

Okay, and more than 10%?

Yes.

Okay, great.

Thanks a lot.

Thank you.

Thank you, ma'am.

Our next question comes from Tim Luke with Barclays Capital.

Please go ahead.

Thanks very much, a couple of quick clarifications if I may.

Was it correct to infer that you were suggesting that your graphics and micro processors are fairly similar rates in the quarter -- in the fourth quarter?

No, Tim, it was the other way around.

CPUs will grow faster than GPUs.

Even though you have the new product in the GPU area?

I'd love them to be equal but that's how I'd handicap it right now.

And just to be clear on Dirk was inferring, he was saying that six to seven is lower end of the norm and that's largely what you're thinking, or you think it's going to be a bit closer to six to seven?

No, I think what Dirk -- what Dirk said is the average is around nine.

This is probably below that -- below the average so I'd call it more in the six, seven range, five to seven.

Thanks, very helpful, and then just on the gross margins.

With all of the things Dirk was outlining on the (inaudible) utilization ASP you didn't think 200-basis points plus of improvement may be achievable?

I'm not going to give that level of granularity, but as Dirk outlined -- as we outline in the current quarter and we outline those are the same factors that we'll continue to make -- we'll make progress on in the current quarter.

Nowhere near the level of improvement we made from second to third, but we will make progress.

And there was some stories about Congo moving from the beginning of the quarter to maybe middle of November, can you just clarify when you would expect to see that launch?

And then also, Bob, if you have any early commentary on the legal backdrop and what we should anticipate in terms of the next three to six months that would be helpful too?

On Congo, we haven't had any change to our original plans so I don't know where those stories came from.

What is that plan just to clarify?

You'll see Congo notebooks out the normal Christmas cycle.

Yes, post October 22nd.

Legal?

Legal, nothing to report from that perspective, so we'll just leave it at that.

We continue to work down the right path.

Trial is scheduled for the end of March.

And then -- Dirk, you said SOI is the 32nm, below that that's when you could start moving the bulk for the micro processors for low power?

No, what I meant to convey, Tim, is that in generations beyond 32nm we evaluate the technology choices as we do in every generation, and we'll give you some more granularity on all this next month.

Very good.

Thank you, guys.

Excellent.

Thank you, Tim.

Our next question comes from John Pitzer from Credit Suisse.

Please go ahead.

Hi, guys, thanks for taking my question.

Congratulations.

Dirk, to the extent a lot of us here on Wall Street are looking forward to the corporate replacement cycle next year, can you just help me understand a little bit better what you've been doing in the corporate client market that might better position yourselves there as that spigot gets turned back on?

Sure.

You said corporate client you're probably referring to the classic corporate enterprise market, big business --

Exactly.

-- 1,000 users and above.

We've got two principal, I'll say -- or maybe three access points in that market.

One is clearly our Opteron business, which is an enterprise-class server offering.

Next would be through enterprise deployments of what are otherwise more like commercial SMB-type platforms, which actually does occur more than you might think.

A lot of enterprises actually don't buy all of the manageability features that are burdened on these enterprise-class platforms so we do participate in big -- enterprise big business with our OEMs on the SMB platforms that we have.

And finally would be our work station graphics business, which is clearly an enterprise focus.

And suffice it to say, Dirk, do you think that with some of the recent successes you're better positioned for this round of PC refresh than you were for the prior?

I would just say well positioned to the extent that the commercial IT sector starts to regain strength that's going to help us.

And then, Bob, I think you said that the AMD design CapEx for the year is still staying at $100 million.

Can you help me understand what Globalfoundries is?

And as you talk about SOI versus bulk, as you guys make your way to 32nm just remind me the timeline and is 100% of the globalfoundry CapEx at 32nm going just to support you guys, or will they also be trying to work on bulk silicon and 32nm.

as well, for potential customers down the road?

Let me dissect a few of the questions.

First on the capital number, the Globalfoundries capital number for this year is approximately $550 million.

AMD, The Product Company is $100 million.

So $650 million AMD consolidated, but those are the two pieces of that.

The capital spend they're spending right now, I'll call it, think of it in a couple buckets.

Bucket one is to serve AMD's needs on additional capacities in 45nm tools, 32nm tools and technology development for 22nm, so the tools associated with AMD's needs to service the micro processor business.

The second bucket is they're trying to get into the foundry business and so they're putting in bulk capabilities at 32nm and below technology nodes so that they can service other people in the industry in the foundry business.

They'll talk a lot more about that, again, at the November 11th conference so those are the two buckets.

And then the third bucket is just an early bucket is breaking ground and building a building and those kind of things in New York.

But those are really the three buckets of the capital spend this year and probably the same story of the capital spend for next year.

And the buckets changing as time goes on.

And just relative to the 32nm ramp for the micro processor business that's still on schedule for first half next year?

We'll be sampling customers in the first half, ramp in production in the second half.

Perfect.

Thanks guys.

Thank you, sir.

Our next question comes from Patrick Wang from Wedbush.

Please go ahead.

Hey, guys, just a couple questions.

First, can you talk about how your 65nm and 45nm mix shifted last quarter and how that's expected to change into next?

I'd kind of characterize this way.

We went from a majority of the shipments in the third quarter being 65nm to much more balance in this quarter to driving to much more significant 45nm from fourth quarter and beyond and then 32nm, as Dirk said, in the second half of 2010 from that perspective of the manufacturing run rates that we'll put into the factories.

Got you, and that's something that's supposed to be a nice margin driver.

Is there any way you can help us frame the impact there?

No, it's pretty hard because to be honest it's fairly complex at the platform level and at the different CPU level.

But clearly 45nm gives you both performance and a smaller dye size and if ASPs are constant, that's a better thing but we actually like to get a pick up of a little bit better ASP new product and a smaller dye size.

Got you, that's helpful.

And then second, I guess maybe, Dirk, I was hoping you could talk a little bit more about graphics.

Given that you feel like CPU grows faster than GPU next quarter could you give us a feel for what you think overall graphics demand is going to be over the next couple quarters, and then maybe follow up with perhaps how you shape up competitively given what we know about your competition out there today?

Thanks.

Yes.

Well first of all, on the competitive front, feeling pretty good about the HD 5000 line up.

First of all, you can buy it today, which is a good step above what the competition recently announced.

I terms of performance and features it's absolutely a killer.

We think it's going to line up very well with the Windows 7 launch here this month.

We're getting really enthusiastic response from our customers as measured OEM design ins, so I think very bullish on that, the graphics market overall.

We're also feeling pretty good about -- we're starting to get some anecdotal input from our customers that discrete GPU attach rates on OEM systems are actually increasing a little bit, even in notebooks, which may surprise some, so overall I see a pretty robust market for graphics, one that we love to participate in because it drives important technology across our platform.

Okay, thanks so much and congrats on a great quarter.

Thank you.

Thank you, sir.

Our next question comes from Jim Covello with Goldman Sachs.

Please go ahead.

Good afternoon, guys.

Thanks so much for taking the question.

Got a little bit of a strategic longer-term oriented question in that I'm trying to figure out how much of Globalfoundries capacity do you need to be in the long run to continue to be able to dictate that product roadmap?

Yes, I understand the question.

We are Globalfoundries only customer today and in even the most wildly successful scenarios for Globalfoundries we are still a big, important, strategic customer for Globalfoundries, particularly given that we will procure technology wafers at the leading edge.

So part of the relationship here is that by virtue of us driving so much volume at the leading edge they're going to be highly motivated -- and I'll tell you they are highly motivated -- to tune their offering to the needs of our products.

For sure, but ultimately they're planning on being a leading global foundry along the lines of a TSM or something along those lines and, again, I just wonder is there any percentage in your mind three to five years out that you think you need to represent of their capacity in order to continue to have the close dialogue on the roadmap?

No, I would say that given the amount of silicon that we procure I expect we're going to have their intention for a long time, not only because of the contractual obligation that they have to our Company but because of the size of the business we represent.

We actually enjoy that sort of relationship with TSMC today given the size of our business with TSMC.

Great.

Okay, thanks so much.

Good luck.

Thank you.

Thank you.

Thank you.

Our next question comes from Uche Orji with UBS.

Please go ahead.

Thank you very much.

Let me just start by asking a little bit of a clarification again on the Win 7 comments with regards to the (inaudible).

The flip side of that, of course, will also imply that the Vista-based PC inventory currently is being drained so if you look across the channel I know you're talking about consumption in Q4 and all that but when you say -- I also know you clarified it as (inaudible) inventory but implied in your comment it sounds like you're worried somehow about inventory in the channel so I'm just trying to get a bit more comfort as to how the interaction between Win 7 inventory and Vista inventory drain (inaudible) to your view of the inventory channel (inaudible)?

Yes, good question.

Again, I'll reiterate.

We haven't heard -- gotten any signals of alarm from either our customers or the channel checks that we routinely run downstream of our customers relative to inventory being anything but in good shape.

With respect to this OS transition, industry insiders vacillate between being worried about an inventory build on the one hand versus consumption stall where people are waiting to buy the new operating system.

As you know, Microsoft has been making available free upgrade coupons on the Vista machines today and that's been helpful in making sure that those machines continue to sell, so I haven't heard anything that causes me alarm.

Rather, we're simply looking at a global world economy that still is not all that rough -- all that robust by many measures and on that basis we're a little reluctant to call return to normal seasonality at this point.

Okay.

A different --

I would actually add one comment and just I like to do what I call my Warren Buffett, which is walk into stores and pretend I'm a buyer and I'm sure every city is different but there's hard to find a computer currently in the stores in Austin, Texas, so I think the OEMs have done a very good job of bleeding down the inventories in anticipation of both Windows 7 launch later this month .

So we feel -- I feel real comfortable the supply chain is as tight as can be and it's prepped for the launch of these new products.

And again, what I think we both don't know, Dirk and I, is what is the consumer appetite to buy new products?

Is it in line with what the OEMs are building or not?

We won't know until we see some of that

Sure, that sounds very -- that's helpful.

A different question.

OpEx, you brought it below the target of $500 million for this quarter.

As revenue starts to improve obviously at some point, bonuses will start getting accrued, how sustainable is this level of OpEx?

Before you even answer that on an ongoing basis, can you just help me understand what we should model for OpEx for Q4?

Well, I'd answer it this way.

We'll use the November analyst conference to talk about 2010, but for the upcoming quarter it'll be in approximately the shoe size of the $500 million.

Okay, and then just my last question.

Your ASPs are up and we did have an environment where PCA fees are down and most components have been up, DRAM is up, (inaudible) started to be weak a little bit recently.

My sense is somebody in some way is getting squeezed, possibly the OEMs, but within that context, how sustainable do you think that the ASPs you have now for micro processors will be, within the context of the declining ASP for the PC and other companies being fairly strong?

Yes, it's a good question and you really have to kind of pick apart the component pieces.

The biggest opportunity that we have to preserve and increase ASPs is to sell our platform offerings into what I'll call a richer mix of PCs, that is higher-priced PCs.

If you look at the overall distribution of PCs -- PC price points across the market we've still got an opportunity to be better represented at higher PC price points, so that's point one.

Your point was really around how much of the bottom cost can microprocessor or the GPU get over time in the face of decreasing (inaudible) prices and hence costs over time.

That certainly is a factor but I think that factor is out weighed by the fact that we're relatively under represented in, I'll call it, mainstream and performance PC price points in the market.

All right, that's helpful.

Do you think that you gained share in graphic this quarter?

I know it's too early to tell, but do you think you gained share in GPU?

I think it's too early to tell, to your point.

Great.

Thank you very much.

Thank you, sir.

Our next question comes from David Wu with G Search -- GC Research.

Please go ahead.

Yes, good afternoon, just two point of clarification.

The first one is, if you're going to stay on SOI for Fusion on your next generation, I guess we're talking about multi-chip packaging with GPU on CMOS and CPU and SOI, would it affect your performance on a combine -- combo chip like that versus fewer silicon CMOS?

That's the first one and I have a question about the thin & light.

I think the retailers are pushing back on the premium pricing that Intel's trying to get on those ultra low voltage products and I was wondering whether your thin & light combo platform is aimed at the same niche, or is it a little lower on the stack in terms of when you talk about thin & light, are we talking about the same comparables?

Good question.

So first, no, our first Fusion parts will be monolithic integration single dye consisting of CPU and GPU technology.

The question about our ultra thin & light offering, we think it's positioned at the right point in the market, squarely between netbooks on the low end and mainstream notebooks on the high end.

And to your point, from what we seen so far in the Intel-based offerings are not appropriately priced or positioned so we think we've got a good opportunity.

Okay.

I assume that we're talking about $600 to $800 on the stuff that you talk about?

We'll have offerings that are lower than that at the platform level.

Oh, I see.

Okay, thank you.

Sure.

Thank you, sir.

Our next question comes from Kevin Cassidy with Thomas Weisel Partners.

Please go ahead.

Thanks for taking my question.

Dirk, you had mentioned in your prepared remarks you thought the IT market was positioned to improve.

I wonder if you could expand on that a little bit?

Sure.

It's hard to put a number on it but the tone of the conversations that we're having with CIOs and other IT decision makers around the industry has certainly changed in the last three months.

Clearly, the wallets are starting to free up, even now a little bit as people perhaps will actually spend the IT procurement budgets that they had at the beginning of the year.

And again, the tone of the conversation about what they're thinking in regards to overall IT spend in 2010 is certainly more bullish than it was.

Yet to be realized yet, but where six months ago people were thinking maybe it's another down year in terms of their IT procurement, not so.

People seem to be thinking they might spend a little bit more money next year, so time will tell.

Okay, so you think there even could be a fourth quarter budget flush?

It's possible.

All right, thank you.

Thank you, sir.

Our next question comes from Hans Mosesmann with Raymond James.

Please go ahead.

Thanks.

Hey, Dirk, can you give us an update on the 40nm as it relates to graphics.

It was a ramp for TSMC earlier in the year and is that now behind you guys in the industry do you think?

Yes, certainly TSMC has made good progress with the 40nm technology.

At this point our new HD 5000 family is only on 40nm and is always the case with the hot new product, early in the launch we're hand-to-mouth on supply.

I expect that situation to improve over the coming weeks and months.

Great and a follow up, more of it from the platform and strategic sense.

The blue team is about to engage in the integration of basically the Northridge in the micro processor and the platform is going to change.

Is there an opportunity for AMD during this transition.

As they go through it are there pros and cons of doing that kind of an approach and when do you guys suspect that you'll be doing the same?

Well relative to integration of the Northridge we in effect did that when we introduced Opteron and Athlon 64 by virtue of integrating the memory controller.

You may be referring to some of the upcoming products the other side has in regards to integrating GPUs in the CPU socket next year?

Yes, that's right.

t the end of the day, customers don't buy integration they buy features and performance, and I feel really good about our graphics technology as compared to anybody in the world, particularly the blue team that you're talking about.

And I think the evolution toward -- away from components and towards platforms and the messaging consistent with that is really in our long term best interest so I applaud it.

Fair enough.

Thank you.

Operator, next question please?

Operator, we're happy to take the next question, thank you.

Sorry, I was having a slight delay on my screen.

Our next question comes from Alex Gauna from JMP Securities.

Please go ahead.

Thanks very much.

I was wondering, you gave some clarification on how you see IT budget spending improving next year.

Can you maybe talk about interest as it falls out around your four, six, eight and eventually 12 core and maybe update us, to the extent that you can, on the timing of some of these multi-core or [magna-core] releases?

Sure.

The headline is no change from what we communicated back earlier this year.

You'll see systems available from our OEM partners essentially very early in the year Q1 timeframe.

The interest from our customers as measured by design ins I would say is good.

The 12-core offering obviously has a lot of benefits in throughput-limited workloads, of which there were many.

The other area where I think we're going to get a lot of uptake is in the HPC part of the market, where bandwidth and flops in each socket is really important and we're looking forward to updating you on some of the other elements around these platforms next month at the analyst conference.

You mentioned HPC, I've had some of my own conversations with CIOs and I've sensed a lot of interest around virtualization in 12 cores.

Is there any difference between some of your multi-core sockets in terms of virtualization interest?

Well, I think to your point that's another workload where our offerings will do very well.

Okay, and you effectively ducked the question on market share.

Am I reading that as, right now you've had a very good quarter, Intel's had a very good quarter, for the best you're able to discern it you would imagine your green competitor is also doing pretty well in this rising tide, or is there something else in terms of mix going on that keeps you from speculating on share with all your 5800 success?

Well, as I said, with the new generation really is going to be more impactful in Q4 than Q3, and beyond that rather than speculate I'll just wait to look at the numbers.

Okay.

In terms of that impact in Q4 you also said you're expecting chip business to grow in excess of GPU, is there some older product lines that are falling down that are keeping that 5800 from driving out performance?

No, I think Bob is just reflecting on what is history in terms of the CPU/GPU mix and thinking out loud, as I agree, using history as a guide is reasonable.

Clearly we've got a lot of opportunity with the new family but as I told you we're also going to be a little bit hand-to-mouth on supply here in the early stages of Q4.

Okay, thanks very much and congratulations on the snap-back quarter.

Thank you.

Huey, we'll take two more questions, please?

Yes, ma'am.

Our next question comes from Craig Berger from FBR capital.

Please go ahead.

Hey, guys, nice job and thanks for taking my question.

Could you just update us on your Fusion plans, what might be the timing on that and any few developments on that front?

Yes.

First, no change from last public communications.

We'll, of course, give a lot of in-depth information in the analyst conference in the next month.

We'll be sampling Fusion technologies first half of next year and ramping in the second half.

Can you talk a little bit about how much of the Globalfoundries joint venture do you guys own, what's your voting share, and what's the implication of either of those for consolidation?

I know you said it's other customers.

And then also as part of that, when does STMicro start ramping, how big might they be next year?

Our governance is 50%, we've got half the hoard seats, our equity ownership on a fully-diluted basis is sitting at about 32%, 33%.

Both of those factors, as well as Bob said, the share of third-party customer revenue from Globalfoundries altogether define a very loosely box of defines consolidating versus non.

And relative to STMicro I'd like to defer that one until we have the analyst conference next month.

Doug Gross will join us and provide that sort of color.

Last question until starting to ramp 32nm you guys are getting deeper into 45nm so it's about a one generation lag, maybe a little less, which seems to be wider than in recent history.

Can you talk about how you plan on closing the gap or what the product or customer impacts might be there as we look over the next few years?

Thank you.

Yes, that's a good question and it's a little -- some of the numbers we throw around give, I think, the wrong impression.

We actually started shipping 45nm CPUs to our customers in Q4 of last year and we started a pretty rapid ramp.

Honestly, the economic crisis really delayed the rate at which we were able to transition our sales from 65nm to 45nm because obviously we had a bunch of 65nm inventory leaving last year that took us much longer to sell out into the market over the course of this year.

So the fact that we're still transitioning to 45nm on an output basis is really misleading relative to when we ramped.

Typically, we ramp and transition in two quarters.

Because of the economy and the reason I said it's taking us longer on an output basis to affect the transition.

This time around I'd expect, as I said, the ramp to start deep in the back half of next year but for that ramp to be more typi -- the transition to be more typical I'd like two quarters.

Thank you so much.

Thank you, sir.

And our final question comes from Adam Benjamin with Jefferies.

Please go ahead.

Thanks, guys.

Just a couple follow ups, obviously on the Win 7.

I think that's been an interesting area of topic for a lot of people.

First off, when did you guys transition over from Vista to Win 7 machines?

Do you mean internal, on our IT house?

No, no, just in terms of when you started shifting over in terms of shipping with machines that you're going to see for the October 22nd release that will show up in stores with Win 7?

So let me repeat.

So you're asking when -- in effect when our OEM customers started to build machines that would be loaded with Win 7?

Correct.

The longest lead time are typically the notebooks that are built in Asia and shipped on boats, so those builds probably started six weeks ago.

Okay.

And then I'm just trying to reconcile, you guys are trying to say, look don't worry about it, we are guiding a little bit more conservatively because we think we may have built ahead for Win 7 as has been speculated by many, but you are looking for the sell through so I'm just trying to reconcile those comments.

If you felt better about it you obviously would have guided better so I'm just trying to better understand what you're thinking in terms of actual Win 7 here?

Well, again, we're guiding a little bit less bullish than typical seasonality because the economy's still got a bit of a ways to go to get back on its feet, number one.

And number two, I personally don't see much reason to signal anything stronger than we said.

I'd rather beat our numbers.

Got you.

And then just last question, just in terms of -- not necessarily your parts but other parts out there.

There's been talk of increasing lead times and potential cancellations and double ordering, in terms of what you're seeing have you seen any constraints in stretching lead times out there for the PC market?

Yes.

Rephrasing your question you're asking about areas where the supply chain is tight on components or lead times?

Exactly, where components, the lead times have stretched and potential double ordering has occurred.

Yes I haven't seen any direct examples of double ordering, double booking yet, it's a good question.

No question that in our supply chain, as an example, across the foundry and outsourced assembly and test area capacity utilization has increased dramatically over the past 90 days and we are seeing lead times stretch out a little bit and that's something our supply chain team is working every day.

Got you, that's all I have.

Thanks a lot guys.

Thank you.

Thank you.

That concludes AMD's third quarter earnings call.

We would like to thank everybody for participating today and look forward to you participating in our financial analyst day on the 11th of November.

Thank you.

Ladies and gentlemen, this does conclude today's program.

Thank you for your participation and have a wonderful day.

You may now all disconnect.