超微半導體 (AMD) 2006 Q3 法說會逐字稿

At this time, I would like to welcome everyone to the Advanced Micro Devices third-quarter 2006 earnings conference call.

All lines have been placed on mute to prevent any background noise.

After the speakers' remarks, there will be a question-and-answer period. (OPERATOR INSTRUCTIONS).

It is now with great pleasure to turn the floor over to your host, Mr. Mike Haase, Director of Investor Relations.

Sir, you may begin your conference.

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

Our participants today are Hector Ruiz, our Chairman of the Board and CEO;

Dirk Meyer, our President and COO;

Bob Rivet, our CFO; and Henri Richard, our Chief Sales and Marketing Officer.

This call is a live broadcast, and will be replayed at amd.com.

A telephone replay will also be available for the next 10 days, starting at 8PM Eastern time tonight.

The telephone replay number is 877-519-4471.

Outside of the United States, the number is 973-341-3080.

The access code for both is 7952575.

I would also like to call to your attention that our Q4 earnings quiet time will begin at the close of business Friday, December 15th.

Also, as a reminder, our next analyst day is scheduled for the morning of December 14th in New York City.

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

Investors are cautioned that our forward-looking statements involve risks and uncertainties that could cause actual results to differ materially from our current expectations, as set forth in the forward-looking statements.

The semiconductor industry is generally volatile, and market conditions are particularly difficult to forecast.

Because our actual results may differ materially from our plans and expectations today, I encourage you to review our filings with the SEC, where we discuss in detail our business and risk factors, setting forth information that could cause actual results to differ materially from those in our forward-looking statements.

You will find detailed discussions in our most recent SEC filings, including AMD's annual report on Form 10-K for the year ended December 25, 2005 and AMD's quarterly report on Form 10-Q for the quarter ended July 2, 2006.

With that, I will turn the call over to Dirk Meyer.

Thanks, Mike.

In the third quarter, our microprocessor unit shipments were up 18% sequentially.

We had unit growth in all three of our product segments, most notably with our mobile processors.

Sales were up 9% from the prior quarter and 32% year over year.

Overall, it was a solid quarter, but I think we could have done still better.

We completed a major transition from DDR1 to DDR2-based products while ramping two new factories and their associated supply chains.

This, coupled with increased customer demand, put an unprecedented burden on our operations.

While we responded reasonably well to this challenge, we learned a lot and will do even better in the future.

Overall, our manufacturing strategy remains on track.

Fab 36 continues to ramp to plan, and we have begun to transition that factory to 65-nm technology on schedule.

We will begin 65-nm shipments this quarter, and will begin our transition to 45-nm technology output by mid-2008.

In addition, we continue to ramp Chartered Semiconductor, and will convert its output to 65-nm technology next year.

Our cost structure will improve as we move to 65-nm technology and complete our transition to 300-mm facilities.

The industry is encouraged by the global antitrust scrutiny of our competitor, and our customers continue to take advantage of the benefits of innovation, choice and growth that we offer.

Examples in the third quarter include -- Dell launched its first AMD64 processor-based Dimension desktop system for consumers and small businesses;

Founder joined Lenovo and [Tonquon] as leading customers of AMD in China with their announcement to launch a full range of AMD64-based desktop, notebook and server systems;

IBM unveiled five new quad-core-ready, Opteron-based mainstream server platforms; and HP and Lenovo launched their first AMD client systems targeted at large enterprise customers.

We're looking forward to launching our next-generation microprocessor core in both native quad-core and dual-core implementations next year.

Our acquisition of ATI continues to progress has planned.

We expect the transaction to close next week, and are pleased to have received all regulatory and shareholder approvals necessary to close that transaction.

In summary, while we are pleased with our results for the quarter, we know that we could have done better.

But despite that, we remain optimistic on both the short term and long term, and are very excited about the opportunities customers are offering to us.

With that, I would like to ask Bob to review the results of the quarter, as well as the outlook.

Thanks, Dirk.

Before I get started, I would like to remind you that as a result of the Spansion IPO, year-over-year comparisons do not include the results of our memory product segment.

Third-quarter AMD sales of $1.33 billion were up 9% compared to the second quarter of 2006, and up 32% compared to the third quarter of 2005, driven by strong demand for server, mobile and desktop processors.

Server and mobile ASPs were up, and desktop ASPs were down sequentially.

We established a new record for total microprocessor shipments, up 18 percent sequentially and up 34% year over year, driven by increased server, mobile and desktop shipments.

Further, customer adoption of AMD Turion 64 processors resulted in a greater than 50% sequential growth and 70% year-over-year growth in mobile dollar sales and units.

Gross margin was 51.4%, down from 56.8% in the second quarter of 2006, largely due to lower desktop ASPs, as prices declined faster than cost reductions, primarily in the desktop area.

However, this cost pressure will be relieved as we move to 65-nm and begin to fully utilize our 300-mm facilities.

Total operating expenses, which include R&D and SG&A, were down $25 million from the prior quarter, better than our previous guidance.

Net income for the quarter was $134 million or $0.27 per share.

These results included a non-cash employee stock-based compensation expense of $16.5 million or $0.02 per share; a $10 million loss or $0.02 per share associated with our 38% ownership in Spansion LLC; and a tax credit of $20.9 million or $0.04 per share, reflecting a lower effective tax rate for the year.

Now, let me turn to the balance sheet.

Capital expenditures were $425 million in the third quarter, as we continued to build out Fab 36 and associated final manufacturing capacity.

We continue with our Lean Manufacturing methodologies for inventory management, and in preparation for a strong fourth quarter, we built inventory by $60 million.

The build primarily comprised of dual-core products.

Days of inventory outstanding at 66 days declined 10 days from the second quarter.

Now, let's turn to the outlook.

AMD's outlook statement is based on current expectation, and excludes ATI operations and ATI-related acquisition charges.

The following statements are forward-looking, and actual results could differ materially, depending on market conditions.

AMD expects demand for its products to be seasonally strong in the fourth quarter and sales to increase sequentially.

Operating expenses for the fourth quarter are expected to be up approximately 10% from the third quarter.

Non-cash employee stock-based compensation expense in the fourth quarter is expected to be approximately $20 million at the current share price.

Our 2006 tax rate is now expected to be between 5% and 10%, down from our prior guidance of approximately 15%.

Capital expenditures for the fourth quarter are expected to be approximately $700 million.

As previously mentioned, we are not providing any guidance today that includes ATI operations.

We look forward to providing you with additional financial details for the combined company at our December 14th analyst day in New York.

At the time, we will provide fourth-quarter acquisition charges and 2007 guidance.

We look forward to seeing you there.

Now, I will turn it over to Hector.

Thank you, Bob.

Less than 100 days ago, at the recent announcement of our groundbreaking acquisition of ATI Technologies, you might remember that I said that we were determined to drive growth, innovation and choice for our customers.

In this recent quarter, I'm pleased to announce that we're approaching this acquisition with a full head of steam on each of these key dimensions.

The AMD growth story continues across customers, geographies and across our product families.

Our third quarter was the 13th consecutive quarter of over 20% year-over-year revenue growth.

AMD Turion 64 processor sales are accelerating, a sign of growing customer demand for best-of-breed solutions in the mobile segment.

AMD Opteron processor sales are strong and growing, a reflection of our system-level power-per-watt leadership and seamless upgradeability to native quad-core in the same thermal envelope.

Our desktop franchise continues to expand, with continued adoption of AMD-based commercial-class systems and AMD LIVE!

PCs, whereas industry analysts' current analysis reported in less than half a year, AMD LIVE!

PC systems surpassed the competition in US retail sales.

We look forward to the launch of Microsoft Vista early next year, a mainstream operating system that will finally allow users to exploit the full potential of AMD64, the industry's first 64-bit-capable x86 platform.

While impressive, we know that growth like this is really an end product, the result of a lot of hard work and a continuous stream of innovation -- innovations like our Torrenza, our open innovation initiative which we announced during the quarter.

Torrenza is our industry's first open x86 innovation platform, allowing our customers and partners to capitalize on the unique capabilities of AMD's Direct Connect Architecture.

World-class customers including Sun, Cray, Fujitsu Siemens, Dell, HP and IBM all announced their support for this unprecedented effort.

Our manufacturing team has been recognized for innovation for many years, and the Fab 36 team in Dresden continues to deliver on their promise to begin 65-nm revenue shipments this quarter.

We're on target to be largely converted by the summer of 2007.

In the meantime, we continue to aggressively grow the world's best and brightest team of design resources, highlighted by the opening of our new Shanghai R&D center and a new advanced microprocessor development facility in Fort Collins, Colorado -- which leads me to choice.

As I have said many times, with our customers who feel completely free to make a choice, innovation and growth inevitably stall.

We are reassured by the courage of our customers to provide real choice to their customers.

AMD continues to offer a better, more customer-centric value proposition for our customers and their customers.

Still, our value proposition to the industry is artificially constrained by the abusive practice of our industry's monopoly.

So, in addition to being optimistic about the continued vote of confidence among our customers, I remain optimistic that global antitrust regulatory bodies will work to create a fair and open competitive environment in the microprocessor industry.

Of course, we believe strongly that our US litigation will result in a verdict declaring that Intel engaged in all of the kinds of illegal conduct that our complaint details.

The microprocessor industry is a growth industry, perhaps one of the most exciting and dynamic in the technology arena.

As an emerging leader in this category, we're dedicated to creating a sustained flow of meaningful customer-centric innovation, creating downstream opportunities, growth and choice for all customers and end users worldwide.

I want to thank the nearly 10,000 AMDers worldwide who are committed and working hard to continue our progress.

In addition, I look forward to welcoming to the AMD family the 4,000 or so ATI employees who will continue their fine work in helping the new AMD create value for customers and shareholders.

I would now like to turn it to Mike Haase for the Q&A.

Thanks, Hector.

Operator, let's begin the polling process, please.

(OPERATOR INSTRUCTIONS).

Adam Parker, Sanford Bernstein.

You said there flat revenue in desktops.

This implies, I guess, your server plus notebook revenue combined grew quite a bit.

Could I assume your server business grew more than 20% sequentially on a revenue basis?

Also, can you comment about maybe did notebook units grow 30% or 40% sequentially or more?

Can you just give us some color there?

No, the numbers are not quite in that mix.

As you heard from Bob Rivet's comments during his discussion, our mobile products experienced a fairly significant growth.

As a matter of fact, it was sequentially greater than 50% for both units and revenue quarter on quarter.

50?

50.

Oh, so you gained massive share in notebook in the quarter, then?

Yes, sir.

We believe.

So that implies, maybe, that you lost a shade of server revenue share, then?

Is that correct?

Based on the numbers that we see, we believe that we did not lose share in the server space, and we grew, of course, quarter on quarter as well as significantly year on year.

But did you grow double-digit revenues in server sequentially?

No.

I'm just a little confused about the gross margins, just because with that kind of revenue growth, I wouldn't have expected your actual gross profit dollars to be lower.

So can you -- Dimension -- I know it was tough desktop pricing.

But were there other moving parts there?

It was just a surprising fall-off to get out of your target business model, despite pretty good revenue growth.

No, I really -- it is mostly characterized by the ASP issue.

Clearly, there's a little bit of cost increase as we continue to produce more dual-core products.

But the lion's share of the issue was the price pressure drop in desktop, which manifested across the whole portfolio on the average ASP of AMD.

Right, but I -- okay, so how do I think about this going forward, then, Bob?

Because you grow your revenue, whatever, $100 million plus sequentially; the operating profit dollars are not up all that much.

So I'm just trying to figure out what the new incremental gross margin is here.

Or how do I think about price stabilization, or how do I forecast your margins, given this kind of pricing volatility?

I think maybe one way to look at it, perhaps, would be, as Dirk mentioned in his remarks, is we have some, we would call it, some transitionary hurdles to overcome in the third quarter, with a transition to a new memory technology, with a transition to a new product, Rev F. All of that places a phenomenal burden on our organization.

Also, frankly, the demand for mobile product was a lot stronger than we had anticipated, and in shifting a rather complicated system of supply chain logistics while doing all this product transitions to got toward more mobile and less desktop, we lost some efficiencies.

We believe that somewhere on the order of 2 of the 5 percentage points that we lost in margin were lost to just not being able to execute as well as we would have liked, as Dirk mentioned.

But the other 3 points, the 60% or so of the margin, is really to a rather precipitous decline in desktop ASPs.

To your point, thinking forward, clearly, we're at a situation where everything we shipped was non-65-nm, so it's the 90-nm variety.

We will ship for revenue, as we stated, in the fourth quarter.

Every day that goes by, we will get bigger and bigger beyond the fourth quarter.

We continue -- we are still not at full utilization of 300-mm.

As a matter of fact, we're not even at half of the productive capacity.

So again, every day will make the wafer costs go down, because we will have better utilization, and the dye costs will go down, due to conversion to 65-nm.

So in what we control, we see a path to continued cost reduction from [last time].

ASPs -- again, we will manage them the best we can.

But that one's the harder one to forecast of the two.

Yes, I hear you.

Was desktop unit up double-digit but decline in pricing down double-digit, or can you give us some kind of balance on --?

Yes, that's kind of in the ballpark.

JoAnne Feeney, FTN Midwest.

Just a question, please, on the memory and the transition difficulty.

It sounds like the switchover to DDR2 and the Rev F were a bit of a challenge.

Are those going to go away in Q4, first of all?

How much of a bump up in the gross margin will we expect from that being eliminated?

First of all, clearly, the product transition complexity that we've dealt with in our operations is behind us.

As I said in the opening comments, we are ramping two new wafer fabs, Fab 36 and Chartered.

In the context of this product transition, and in the context of increased customer demand, resulted in us having a hard time matching that customer demand to supply availability in the weeks the demand was really there.

So, in aggregate, we did a pretty good job operationally.

On a week-to buy-week basis, we didn't always have exactly the mix of product that our customers would have wanted, which is clearly not good from our perspective as a customer-centric organization, and operating that way doesn't allow us to maximize the value we can draw out of the business.

Those issues are largely behind us.

In the server space, your competitor is claiming to have been quite well there, too.

Are you seeing particular increase in competition in one area of the server space, 2P versus 4P, for example?

There's no question that if you look at the technological capabilities of our product, we're by far the most differentiated in the 4P category.

As a result of Intel's announcement of their recent server product, we are facing a bit more competition in the 2P category.

So I'd say that's a fair characterization.

Michael Masdea, Credit Suisse.

What I want to clarify is the strategy on the desktop side.

It sounds like you're talking about cost reductions -- help me out there -- but your competitor is being a little bit more aggressive with their manufacturing plant.

What's your longer-term view, especially in terms of trade-offs between market share and profitability?

First of all, relative to desktop, we're going to serve the demands that our customers expose to us.

Over time, I would expect that as we grow into a broader part of the marketplace, namely commercial, we've got a huge opportunity to grow our gross margin pool in aggregate.

As Bob said, the ASP issue is a little harder to have direct control over, because we respond to a competitive environment.

We do so in a way that allows us to make our relationships with the customers we care most about broader and deeper, while at the same time managing profitability.

Understood.

So, on the server side, maybe that's a place worth spending a little bit of time.

Where is your kind of customer -- or, excuse me, on the enterprise side, in the mobile and desktop, where is your customer traction now?

Where do you expect it to be about a year from now?

As you have noticed, we're continuing to add not only customers that are playing in that space, but within the customers and we're already doing business with we're seeing a broadening of their product portfolio.

So that alone is enabling us, looking forward, to hope to gain share in that particular segment of the market.

We've clearly seen strong demand on the mobile side, and you now have on the Turion platform enterprise-class mobile platforms that we did not have last year.

So I expect us to grow both desktop and notebook, but Turion is definitely proving to be a very valid platform for the enterprise market, with all the new products that have been introduced, frankly, in the first and second quarter are now coming to life.

The last quick one for me is really on the capacity side.

If you listened to the call on Intel last night, and listen to you guys and look at your forecast, it seems like both companies are building capacity in the future based on market share -- at least, gain for you and at least holding for Intel.

How does that ultimately play out?

Why is that not going to be a problem for the overall pricing dynamic for many years to come?

Is there some point when someone yells "Uncle!" and pulls back a little bit?

We have outlined before that our strategy on capacity has been to continuously and steadily gain share over time.

We've engaged -- we have gone from a significantly skewed profile of our customer base a few years ago, towards distribution and consumer, to one that has shifted over the years rather significantly towards OEMs, particularly Tier 1 OEMs, and beginning to make inroads into the commercial space.

We carefully planned that out with our customers, and we planned it out over a period of more than just a quarter or two, an outlook of three years or so.

Based on that, we put our plans for capacity in place.

So far, every capacity plan that we have and that we have outlined has been based on input from our customers, those that we have had for quite some time in addition to the newly acquired ones, to continue to execute to that.

The plans that you have heard us talk about is the expansion of Fab 36, the conversion of Fab 30 to 300-mm, the use of Chartered, as well as the potential new factory in New York are all based on all those plans that we have collected from our customers.

So we don't base it on just a guess on our part and hoping we can sell it later.

They are really carefully thought out in conjunction with our customers.

That makes sense.

I guess, just a really quick follow-up, it would be -- even if you have that environment, you take that share, is the extra capacity in the industry going to cause a problem and take away -- drives into a profitless kind of prosperity environment for the industry?

What keeps that from happening if Intel has all this extra capacity?

Here again, I would just reference and reiterate Hector's comment that we are putting capacity in place based on what our customers tell us their plans for us are.

There's no question, as we expand in the new segments -- example, commercial desktop -- we will introduce price competition into those segments where previously there was none, because those segments were dominated by a monopoly.

We are not afraid of that, and in fact, that's our strategy.

Glen Yeung, Citigroup.

I guess this is a question for Dirk.

You saw a lot of inventory build in the quarter, and I recognize at the beginning of the quarter you said you wanted to do that.

But I wonder if the inventory build was exactly what you wanted, particularly given that you referenced you thought you could have done things better during the quarter.

Actually, most of the inventory build that we did was WIP that is being built in preparation for Q4 sales.

The issue within the quarter wasn't so much one of building inventory of finished goods that we couldn't sell, but rather having the right product and the right mix available for the right customer at the right time.

If I can to add to it, not to argue with the numbers, but I think our days of inventory went down by 10 days quarter on quarter.

So we don't see any of the WIP that Dirk referred to as being anything other than strategically positioning the product for what we believe will be a strong seasonal growth in the fourth quarter.

So, assuming that you are better able to manage the mix in Q4, should we assume that inventories then start to go down?

Yes.

Our normal seasonal pattern would be inventories would drop.

We would drain inventories and fourth quarter.

But again, our inventories are based on our projection of continuing to service our customers.

We continue to acquire more customers, and we continue to ramp Fab 36 and Chartered Semiconductor.

So I am actually hoping that we will build inventory a little bit to go out of fourth quarter in preparation of 2007.

It begs an interesting question about gross margin.

I think you talked about some efficiencies that were worth 200 basis points in the third quarter;

ASPs, which were worth 300 basis points in the quarter; and now we have got potentially rising inventories in the fourth quarter, or hopefully rising inventories.

Assuming that inventories go up, assuming that the inefficiencies disappear, we're already talking about 200-plus basis point increase in gross margins.

Intel is suggesting that pricing is not as bad.

So is it fair to say that under that assumption that pricing is not that bad, that we could see a greater than 200 basis point, maybe greater than 300 basis point increase in gross margins?

Your math logic is fine.

The only part I would counter with, that I'm not sure you got it right -- I said we would grow less inventory in fourth quarter than we did in third.

I'm trying to grow inventory in fourth to be positioned for 2007.

So it's actually less than what we just did.

But your math is not too far off.

What is normal seasonal growth for you in the fourth quarter?

Just what is your interpretation of that?

Market conditions would say this over, I call it, a recent history of time.

That is, on the low end is 6, on the high end is 13, and most likely averages around 10.

I think this was kind of asked, but I'm not quite sure I caught the answer.

If I would think about 2007 just broadly, your strategy in the year, what is your position on share versus margin?

I don't think we look at our business from that point of view.

We look at it more as what do our customers want?

We are very customer-centric, and we go from there.

Overall, at a pretty high level, I would tell you that it appears that we are on track this year to grow units somewhere on the order of 35%.

Our plans for next year is to grow units approximately the same amount.

And to the margin question, we still -- while, this current quarter, we bounced for the first time in a long time below our zone, I'd call it, of 55% to 60%, that's our goal.

Our goal is to live in that zone.

Obviously, every quarter is dependent on the situation that appears at hand, and we will deal with it accordingly.

But our long-term model is to try to get into that 55% to 60% zone for our microprocessor business.

Tim Luke, Lehman Brothers.

Just a question with respect to the pricing environment, in terms of the linearity of how you saw that as you moved through the quarter.

Did the environment, in terms of ASP movement lower, get more challenging towards the end of the quarter?

As I think Glen mentioned, Intel has talked about some firming, potentially, of the ASP environment going forward.

Is that something that you might anticipate, or do you think it's somewhat premature to consider that?

I think the third quarter was largely impacted at the beginning of the quarter by the ripple effect of Q2 and the movement of the competition then.

I think, when the middle of the summer was behind us, we saw some stabilization in the marketplace.

It remains a very competitive marketplace, but I would agree that there has been some stabilization in the second half of the quarter.

So perhaps a decline in gross margin, given that ASPs are broadly in line, perhaps, with your expectations, related to what you are referring to, in terms of this mix and match, in terms of what your customers wanted and what you were able to deliver?

Maybe you could expand a little on that mix and match point you made?

I'm not sure I got the question.

You mentioned the ability to match demand with what you had available as a factor influencing your margin, or perhaps you were relating to the efficiency.

Perhaps you could just expand on what you meant by that?

Well, I think one of the things that happens, as we get more skewed towards having a higher mix of Tier 1/Tier 2 customers, we have a better understanding of with expectations are from them in terms of forecasting.

But frankly, one of the challenges that we had in the third quarter is that a lot of the people that are involved in the go-to-market channel of retail, for example, did not anticipate such a rapid move towards mobile, just as an example.

Therefore, when you take roughly, in round numbers, 100 days from the time you get an order to the time you can actually produce it, it is very challenging environment to try to adjust mix within a quarter.

So we are learning from that in the third quarter.

We spend a considerable amount of time with customers anticipating the fourth quarter being strong, trying to understand their needs better.

We think we have done a better job, as Dirk pointed out, in anticipating what we're going to need to do this quarter, and we expect that mix to be better managed, and make our customers a lot happier this quarter than we did last quarter.

Just to be clear, in terms of the revenue, Bob, when you're saying seasonality, the average is 10%, so stronger-than-normal seasonality would imply that your target would be above 10% revenue?

Is that how we should basically figure that?

That's a good characterization.

We'd say average is 10.

Lastly, just on your OpEx, it went down a lot in the quarter.

Now you're guiding it up 10%.

Could you just give us some color on that volatility there?

Clearly, always, fourth quarter is a heavy consumer quarter, a lot of merchandising and I'll call it market development funds are associated with that.

So I'm signaling Christmas is going to happen.

We're going to have that kind of activity from that perspective.

Then, on the other side, we continue to expand these design centers we just opened.

We're trying to ramp and hire the appropriate engineers to continue to accelerate, keep our roadmaps where they need to be and accelerate roadmaps and add new things to the roadmaps.

So it's a combination of both.

We were a little bit Hunter, to be honest, in the current quarter.

So, it's kind of balancing your way through it.

Joe Osha, Merrill Lynch.

I'm wondering if I can drill down a little bit on the manufacturing output.

Can you give me some sense roughly for the current quarter, and then for the fourth quarter, what portion of your output is coming from 25 (indiscernible) Fab 30 versus Fab 36 versus Chartered?

Obviously, units you probably don't want to talk about, but maybe just in 200-mm equivalents?

I can certainly give you a feel for where the wafers are coming from, from our internal factories.

We started ramping Fab 36 in Q4 of last year, and it's roughly a two-year ramp to get to the 20,000 a month point.

We're on a schedule that says that we are at about the 10,000 a month output level now.

And that's 300-mm.

That's 300-mm you're talking about.

That's right.

It's 300-mm.

Fab 30 is really pretty stable at -- now I'm going to have to go to 200-mm starts a week (multiple speakers).

So it's around 28,000, 30,000 a month?

28,000 to 30,000 200-mm.

We don't, I think, want to be to specific on how we are loading Chartered at this time.

But it's a much smaller number, yes?

Very much smaller.

Now, are you all still thinking that -- I knew you talked about Fab 30 maybe being kind of incrementally brought down, or at least partially brought down and converted to 300-mm starting in late 2007?

Is that still on the table?

Actually, we will start taking Fab 30's 200-mm output down in about the second quarter of 2007.

I'm sorry?

Second quarter of 2007?

2007.

Then, we will be getting 300-mm output back out of that facility in very early 2008.

Will Fab 30 continue to produce for that?

Will it completely go away, or --?

The total utilization of the floorspace does not really dip any lower than about 40% over the course of the transition.

But I could see Fab 30 drop to, say, 10,000 -- 10,000, 12,000 a month while you were in the midst of this, yes?

A quarter.

I'm sorry, a quarter.

Absolutely.

Yes.

No.

Yes.

Just to remind everyone, we built an additional facility on the site to kind of deal with I'll call it the transition facility.

So there will be a dip, but we're not actually shutting down the facility.

We continue to run it.

We will take it down a bit, transition to 300-mm and, in 2008, be completely flipped over to a 300-mm facility.

Again, roughly the same size as Fab 36, in the 20,000 wafer per month level.

In 2008?

In 2008.

Bob, while that fab is being converted, how does that work financially?

Does that create any kind of unabsorbed overhead or period costs that are not correlated with output?

No.

We think, between the needs of Fab 36 and the needs of Fab 38, we're going to call it, we will actually have the appropriate staffing.

The tools on the 200-mm are almost all fully depreciated, anyway.

So to some degree, we will actually be able to -- and the chemicals and gasses, of course, will flex down with the appropriate volumes.

So we actually feel like it's not actually a big cost penalty to do the transition.

You mentioned that Fab 38 -- I'm sorry, Bob.

Since you kind of put that on the table, can you give us (multiple speakers)?

Fab 30 is going to be renamed and called Fab 38.

That was part of a strategy to actually get more grants and subsidies on the new toolset going in there, because we are really recasting to a new facility, a 300-mm facility.

So, no comments on New York State, then?

That's still in the plan, as Hector said.

But in the sequence, it's Fab 36, Fab 38, New York.

One thing that you were alluding to, I think, and I just want to make sure you understand, is that the combination of these conversions, as well as the 65-nm that is occurring as we speak, and the ramping-up of Chartered leads us to plan our capacity next year.

So, while Fab 30 has a dip in the wafer starts in 200-mm, the overall unit output throughout the year increases.

So we expect next year, as I mentioned earlier, to still be approximately 35% up in units from this year.

Cody Acree, Stifel Nicolaus.

Maybe a point of clarification.

The 200 basis points of inefficiency -- do you expect to fully get those back in the fourth quarter?

Is that a ramp [of fleet] back in over the next few months?

We expect to fully get them back in the fourth quarter.

Can you talk about the differential in gross margins?

As you have talked about the Tier 1 and Tier 2 higher mix of revenue, is there anything meaningful in that gross margin line that we saw this quarter, as you have had more success at Dell and some of these Tier 1 OEMs?

No.

I think that it's more complex than that, and I think it would not be productive to go into it.

But just to give an example, a customer that buys an awful lot of product is more likely to get better pricing than a customer that buys a lot less product.

So it's much more complex than just saying Tier 1, Tier 2 and channel.

It's a function of mix, in other words.

Yes, it's mix.

The only other color I would add -- I want to remind you the strategy is still server, mobile, desktop in a priority sequence.

Gross margin will fall in that direction, too.

Although desktop and mobile at moments in time will be equal, they will still continue to flip positions.

But clearly, server is where the higher gross margin is.

We see that today and on a continuous basis.

Average revenue growth in the fourth quarter, around 10%.

Are you allowing any assumptions for Vista for Q4?

What would you expect as we push into 2007 for normalized Vista impact?

We are not prepared to talk about 2007 at this point in time.

We will do that at our analyst day on December 14th.

For the fourth quarter, are you allowing anything for a delay of purchases because of Vista?

No.

We don't see this to as a negative issue in the fourth quarter.

[Sandeep Panjukhar], Morgan Stanley.

It's Mark Edelstone.

I had a couple questions.

The first one, Hector, was just to get a sense for how big the OEM business is for you now.

So can you give us a sense as to what percentage of your units is coming out of OEM, and how you expect that maybe to trend here as you go forward?

I am not going to be too specific, but I will tell you that it has flipped from what it was four years ago.

It is more than 50% today.

I think we're beginning to approach a stable mix, give or take a little bit.

So, we're pleased with the position that we have and especially, of course, with the latest customer acquisitions.

I think that gets us to where we would like to be.

Just as a follow-on, using the third quarter, could you just give us what maybe the percentage breakdown would look like on units between notebook, desktop and server?

I don't think we're prepared to give that kind of detail.

Chris Danely, JPMorgan.

You talked about gross margins popping back up in Q4.

Can you just give us a sense of the trends in startup costs and depreciation going into 2007?

Can we expect those to increase with the ramp of 65-nm and the CapEx going up?

Break them into two.

CapEx, as I signaled last time -- I didn't talk about it this time, but therefore I didn't change it -- is a large number next year, $2.5 billion.

That is the continued, as what Dirk said, to fill out the equipment set in Fab 36 and start the conversion of Fab 30 to 200-mm facility to Fab 38, the 300-mm facility.

And then, of course, the associated back-end capacities needed to turn that die into finished goods.

So capital will continue to be at a fairly high rate next year as we build out capacity.

From a cost standpoint, we are, as I said to Joe Osha's comment, we believe there's not a huge penalty in the conversion of Fab 30 to Fab 38, and we're actually paying most of the costs today to develop the technology and put the 65-nm into production.

So that actually get better as time goes on, not worse, from that perspective.

On the gross margins, I remember last quarter you guys said you don't want to participate in any -- I think it was deeply discounted opportunities.

But it seems like you might have participated in some of them in Q3.

Can you just talk about that?

Is that a true statement?

That's a fair assumption.

As Dirk was pointing out, we completed a transition from Rev E to Rev F. As we did that, we felt that it was appropriate to be a little more competitive at the low end of the market, as we completed that transition.

But we're going to continue to be very selective going forward in the business that we participate in.

It sounds like pricing got a little better at the end of the quarter.

Can you just talk about why it got better at the end of the quarter?

Do you anticipate pricing to be, I guess, competitive in Q4, or getting a little bit better because it's a stronger seasonal time period?

Well, as Einstein always said, everything is relative.

After the disaster applied by the competition in the industry in the second quarter, effectively Q3 was somewhat mild.

Going into the first half of next year, why wouldn't we see the price wars crank up again?

How long or how low would you let profitability go before you would start to pull back, or will you just keep going and try to gain market share?

Here again, the way I would think about this is the market is not a homogenous blob; there are segments.

Where we have historically competed or been present in the market, we have always had price competition.

As we have had the opportunity to compete in new areas, we have brought price competition to those areas.

So, from the AMD perspective, our share growth doesn't imply increased price competition on our business, because we have always had it.

Does that make sense?

Yes, I guess.

I don't remember the last time you guys had 51% gross margins in processors, so my concern is that we could go lower in the first half of next year.

Does that make sense?

No.

I'd say I think that's being fairly pessimistic.

That's saying we wouldn't execute 65-nm, because 65-nm is a fairly steep cost reduction that takes place, and continued utilization of Fab 36 to fill it to the 20,000 wafers per month.

So in what we control, we feel like we have, clearly, a great path to pretty sizable cost reductions in our manufacturing capability for our output.

Prices, to me, is a market-dependent issue.

A question that came up before -- we think Vista is a good environment, not a bad environment.

So, to some degree, 2007 is -- it doesn't look like a bad year.

It looks like a good year.

So we believe we're in pretty good position, again, to get back to the zone we want to be in, which is a gross margin of 55% to 60%.

That makes sense.

David Wong, A.G. Edwards.

[Further on] the question of inventory and your inventory plan, will you be required to build [the bare] amount of finished goods inventory [staged] for customers that have just-in-time manufacturing in the future, or is your inventory profile, as it is at the moment, likely to remain unchanged?

All our customers have different requirements to have inventory staged or in their hubs or whatever.

Clearly, as we continue to acquire new customers, those dynamics continue to change.

I would say probably the big OEMs do like more product to be staged for them, so they can pool it as we see there their market demands move around.

But that's also why we have Lean Manufacturing type thought process and organization put into the system, so that we can keep it as lean as possible, but still servicing what the customer wants.

The bottom line of that is we're maniacally focused on inventory management.

We will grow it where we see necessary strategically, but to make sure it's fresh, and just as much as we need to run the business.

So you won't have to do a significant inventory build in the December quarter to meet your customers' needs, then, for staged inventory?

Correct.

Krishna Shankar, JMP Securities.

As you look at two of the areas with the highest gross margins you have had in the past, sort of the high-end enthusiast and gaming PC market, and also servers, it appears that the competition is more aggressive, and perhaps has a better product lineup there.

Can you outline your product plans to sort of regain the leadership in the high-end PC gaming market, and also regain your leadership in DP servers?

There's a lot being written and said about the high-end gamers market.

We actually saw a double-digit increase in revenue in that segment in the third quarter, despite all the hype coming from the competition.

So yet again, I want to insist on the fact that it's the platform that matters in that market, not just the core performance.

From that perspective, we're going to launch, in the next few weeks, a new platform that for the first time was built from the ground up with the needs of the gamers in mind.

Not only the processor needs, but basically the cooling needs, the power supply needs and the space on the motherboard needed to have a world-class platform.

That platform will be out of the chute, optimized for Vista operation.

As far as the server market is concerned, the strength of the Opteron architecture remains intact, as demonstrated by our continued growth in that segment and the particular strength of our 4P offering.

I think Fortune 500 companies that are looking at the two offerings are recognizing that, although the competition has improved their relative performance, they still have not answered the question of what is their future architecture, and they are shipping obsolete products.

As a follow-on, can you also outline your relationship with NVIDIA with the ATI deal really closing now?

Can you talk about the relationship with NVIDIA going forward?

We're standing for open and fair competition and platform innovation.

I'll give you an example of that.

Our launch partner for our upcoming gamer platform is NVIDIA.

So our relationship with NVIDIA is excellent, and we're very happy to have them as a partner.

Jim Covello, Goldman Sachs.

A lot of good questions have been asked, so I just have one big-picture question.

If I take a step back over the last couple of years, you've done a terrific job executing versus your plan and gaining share.

Over the last couple quarters, though, last quarter was the first quarter in a while where we had a revenue miss.

This quarter is the first quarter in a while where we had a margin miss.

As we head into next year, the bar for execution goes higher, because of the debt you are bringing on and because of the incremental depreciation.

How do you get comfortable that you can get back to executing versus the plan when, admittedly, some of the things like ASPs are kind of out of your control, and your competitor now has a little more incentive to try and hurt the market a little bit, because of the debt and the depreciation that you guys are ramping?

How comfortable can you feel around that?

Well, first of all, we're very confident in the roadmap that we have, which we continuously share in intimate detail with our customers.

We are excited about that roadmap and we are executing to that.

As Dirk pointed out, next year we're introducing products that are going to continue to enhance that.

In terms of managing to the [bumps] relative to ASPs that are artificially created by an event of the marketplace -- such as a competitor perhaps building way too much inventory, being late on their products and therefore having a lot of old products to unload -- frankly, we don't agonize over that.

We don't control that.

We felt terrible in the second quarter that that was the way it was handled, but we had complete confidence that, over the long run, that would not be a sustainable situation. (Multiple speakers).

My question, if I can interrupt -- unfortunately, the shareholders and stakeholders do agonize over that, because you're leaving yourself not as much room for error by ramping the debt and ramping the depreciation.

So I understand it's a long-term plan, and you have had an excellent roadmap, you continue to have an excellent roadmap.

But that hasn't prevented outside influences from causing you guys to have issues the last two quarters.

So, how can we know that those outside influences are not going to have continued influences over the next couple quarters, given that the bar is going higher?

Well, I think we are credible in the things that we tell you we are doing.

One of the things we are is, as we have gone through this challenging couple of quarters for different reasons, we have become better at learning how to deal with those.

We are better planners.

We have a better idea of how to anticipate and move in those directions that are necessary in the marketplace.

But the best answer I can give you, frankly, it doesn't come from me; it comes from our customers.

Our customers have spoken.

You have the products that IBM has launched that are outstanding to penetrate the enterprise, are based on our products today and future existing products, because the enterprise, in particular, the products are quad-core-ready.

You have one of the fastest-growing companies in the server segment today, which is Sun, is totally committed to our product line.

You have a new customer that just signed up named Dell, who is also committed to working with us going forward.

Frankly, we're not through with customer acquisitions.

We still have a number of customers around the world that have yet to sign, and we are confident that in 2007, we will see additions to that, too.

So yes, I recognize and I admit that there will be some short-term shareholders who are going to experience some concern and agony over a quarter or two where things don't go perfect.

But we're confident, and we're hoping that we can convey that to you.

We already conveyed that to our customers, and that is why they are signing up to uses long-term and in big ways.

Doug Freedman, American Technology.

If you could comment a little bit on any impact that you might have seen over the DDR1 pricing rises in the marketplace, and what impact that might have had on your business?

And if you could give us some insights into the percentage of products that are presently supporting DDR1 versus DDR2?

First, the DDR1 and DDR2 prices have been kind of switching back and forth position over the course of the quarter.

I don't think there's a real trend there; it's very spotty.

Frankly, my insight in the market tells me that NAND has a bigger impact on DDR pricing than anything else, because a lot of the switching back and forth on the line is between Flash and DDR memory.

As far as the second part of your question, there's still a large portion of the server sales that were executed in the third quarter that are based on DDR1 technology because, as you know, the cycles on those projects products are longer.

So, even though there has been a lot of announcements from all our key partners of Rev F-based products, the reality is that those products will not hit to market until Q4 and in some cases Q1, and that's all per the plan.

For the client side, it's essentially converted.

Going forward, we are shipping DDR2-enabled products almost exclusively.

We have talked quite a bit on the call so far about the Fab 38 plan and expansion and moving into 65-nm.

Would I be correct in believing that Fab 38 will be your first 45-nm facility?

Any thoughts on how quickly we will be able to move from 65 into 45 nodes?

Our plan is to begin to ramp 45-nm products from an output perspective in roughly mid-2008.

For the current plan, we will actually start to do that more or less simultaneously between Fab 36 and Fab 38.

Ross Seymore, Deutsche Bank.

Just one quick question, returning to the OEM versus Channel side of the equation.

You talked about the OEM business getting up to now greater than 50%.

Is there any difference in the margin structure, either on the gross margin line or with some sort of co-marketing dollars below COGS in the SG&A line, that would create different profitability for the customers on the OEM versus the channel side of the business?

As we indicated, we have the long-term plan to stabilize our business by having a greater percentage of it derived from OEM.

But even within the OEM, of course, there's large marketmakers that are very important, in order to secure long-term visibility on the market.

We have made a lot of progress in that direction.

Now, these customers, as Dirk was pointing out, are participating in various market segments.

Typically, they have a wider product range, and participate in many more segments and in some cases many more geographies.

So, again, there's no simple answer to your question, because it's a matter of mix -- mix of products and, of course, mix of geographies and, within those, mix of commercial versus consumer segments.

So overall, it's very clear that the largest customers in the world want to command a premium over the smaller customers in the world.

But the reality is that, again, you have to go case-by-case.

As far as the cost of supporting those customers, frankly, we put dedicated teams on all of our key accounts, but we also have dedicated teams in the channels.

So there's not real large deltas in terms of cost of support between OEMs and channels.

By the way, the channel continues to be -- and will be for a long, long time -- an incredibly important part of our go-to-market strategy.

Over the time, Henri and his team have reduced the complexity of working with the channel, and we have now what we believe is a very strong position in that space.

When we are growing units, as I mentioned earlier, 35% plus a year, we have an opportunity for those channel partners to still grow, despite the fact that the mix continues to change towards the OEM.

Again, in that balance, how do I think about the substitution effect possibly kicking in, where your OEM business is ramping very nicely -- we're all well aware of the Dell side of things.

I would imagine some of that is going to be targeted to areas where you already have current success.

That 35% unit growth you are talking about -- how do we get our arms around what sort of substitution effect might be impacting you within that 35?

I think, first, to assume that one customer ramping is automatically going to cannibalize existing business is a place where I'm not right now.

It's too early to tell, anyhow, how things are going to shake out from that perspective.

Again, go back to Dirk's point -- we are participating in a lot of new markets, whether they are geographies or segments of the market.

So that's where most of the growth is going to come, and we're very careful in working with our partners to position each of the new platforms that we're bringing to market to maximize our opportunity.

Fair enough.

Then, the final question, to go back to a topic that has been hit on a couple times.

You have mentioned about your customers dictating the expansion in your capacity, and you are building along the lines of their desires.

Given the capital constraints that that might put you under -- what sort of reassurance or assurances can they give you on the pricing side of the equation?

I would assume there's got to be a point at which they want capacity, but they also want the lowest-priced microprocessors they can get.

So they have to have a balance with you being profitable.

Help us understand how that trade-off works.

Well, I am not aware of any plan that we have built with any customer or any partner that, over a period of time, does not show a reasonable return to our shareholders and is financially sound.

If that were not the case, we would not be able to do it.

Fair enough, thank you.

Great, thanks.

We will conclude the call.

I look forward to seeing everyone at our analyst day on December 14th.

Thank you.

This concludes Advanced Micro Devices' third-quarter 2006 earnings conference call.

You may now disconnect your lines at this time, and have a wonderful evening.