超微半導體 (AMD) 2004 Q4 法說會逐字稿

Ladies and gentlemen, thank you for standing by and welcome to AMD's Q4 '04 earnings conference call.

At this time all participant lines are in a listen-only mode.

Later, we will conduct a question-and-answer session and instructions will be given to you at that time.

If you should require any further assistance, please depress star, and then 0, and we will assist you off-line.

And as a reminder, this conference call is being recorded.

I would now like to turn the conference over to the Director of Investor Relations, Mike Hasse.

Please go ahead, sir.

Thank you.

Welcome to AMD's fourth quarter earnings conference call.

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

Bob Rivet, our Chief Financial Officer; and Henri Richard, or Executive Vice President of Worldwide Sales and Marketing.

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

The telephone replay number is 800-475-6701.

Outside of the United States, the number is 320-365-3844.

The access code for both is 765351.

A telephone replay will be available for the next 10 days starting at 7 p.m.

Pacific Time tonight.

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 current expectations.

The semi-conductor industry is generally volatile and market condition are difficult to forecast.

Because our actual results may differ materially from our plans and expectations today, I encourage to you review our filings with the SEC where we discuss in detail our risk factors and our business.

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 28th, 2003 and AMD's quarterly report on Form 10(Q) for the quarter ended September 26th, 2004.

With that, I'll turn it over to our Chairman, President, and CEO Hector Ruiz.

Thank you, Mike.

You know, the fourth quarter of 2004 was, for AMD, a solid quarter in microprocessors, but frankly, a freakin' dismal quarter in Flash.

In processors we showed some strong fundamentals, but in my view, despite dramatic year-on-year gains, it was still an underwhelming end to an otherwise break-away year.

It is worth noting that in 2004, we grew our processor business by a commendable 29 percent year-on-year.

And based on the strength of our AMD Opteron franchise, we continued our unprecedented growth in the commercial server and enterprise segments.

And at the end of the year, over 40 percent of the companies in the Forbes 100 are satisfied AMD64 customers, a very strong sign over our possibilities for expansion in this year.

AMD64 base processors now represent one half of our processor business.

On this core and a combination of expanding demand, continued strong yields, and world-class manufacturing execution.

Our Athlon 64 FX processors held the performance ground from wire to wire through the course of the year, almost unheard of in this business.

And our new AMD Sempron processor family, in only 6 months since its introduction, has become a key part of our successful growth into high-growth markets such as India, Russia, and China.

In the fourth quarter, we saw record server and mobile processor sales and double-digit percent unit growth across all of our segments.

Yet despite these successes, we remain unsatisfied with our performance in processors, where we believe we're not capturing yet our fair share of the overall opportunity that we see before us.

And we will address this shortfall effective immediately.

In our Flash business, we had an awful quarter.

In the context of an aggressive pricing environment and a continued supply and demanding balance we suffered from poor execution in Japan and delayed product qualification in a wireless segment.

Yet, overall, we begin the new year in Flash in much better shape than a year ago.

We completed the successful integration of the AMD and Fujitsu businesses.

Customer acceptance of our revolutionary MirrorBit technology continues.

And we believe we're set to revolutionize the Flash memory business with our ORNAND architecture, providing the best of in NOR and NAND technologies in a single device.

In short, while the fourth quarter was disappointing end to an otherwise good year, we believe we are better positioned in both of our businesses than we have ever been.

And at this point, I want to ask Bob to review the financial results for the quarter as well as the outlook.

Thanks, Hector.

As detailed in our press release, we reported record fourth quarter sales of $1.264 billion, up 5 percent compared to the fourth quarter of last year and up 2 percent compared to the third quarter of 2004.

Key highlights in the quarter were a 9 percent sequential microprocessor sales growth, record sales in server and mobile, and record desktop unit shipments.

This helped drive all-time high quarterly sales and our tenth consecutive quarter of increased gross margin percentage.

Operating income for the quarter was $20 million, and declined from the third quarter due to the loss from our memory business.

This performance represents the fifth consecutive quarter of profitability.

Loss per share on a GAAP basis was $.08.

Included in this were charges of $.13 per share, largely associated with the conversion and retirement of debt, thereby allowing us to remain profitable on an EPS basis.

Gross margin improved to 41.2 percent for the quarter, above the 40.5 percent reported in the third quarter, and a solid 5.8 percentage point improvement from the fourth quarter of last year.

This improvement was driven by a combination of continued improvements in our CPG business, the mix of CPG business as a percent of our overall business, offset somewhat by a reduction in our memory business gross margin.

Research and development spending was $253 million for the quarter, up 9 percent from the prior quarter.

The growth was driven by Fab 36 start-up cost, as planned, and other server and work station -- product development efforts.

In December, we started placing equipment in Fab 36.

Fourth quarter marketing, general and administrative costs increased 21 percent as compared to the third quarter of 2004, due to seasonal merchandising activity, long-term branding investments and first-year SOX certification costs.

Cash flow from operations was $362 million for the quarter, up 37 percent from the third quarter of 2004.

EBITDA was $317 million for the quarter.

Now, I'll switch to the business overview for the quarter.

I'll start with our Computation Product Groups.

CPG sales were $730 million in the quarter, a 26 percent increase over the same period a year ago and a 9 percent improvement from the third quarter.

Unit sales in the quarter increased sequentially by more than 15 percent, with double-digit growth in each of the product segments.

ASP and our AMD64 server and mobile increased.

However, overall ASP decreased as compared to the third quarter, driven by desktop demand in high-growth regions.

Unit sales of AMD64 processors grew by more than 70 percent sequentially, and now represent 50 percent of total CPG dollar sales.

We saw continued demand for our mobile processors, and we shipped record server and desktop units in the quarter.

CPG's operating income in the quarter was $89 million, or 12 percent of sales, flat from the third quarter.

CPG's gross margin increased in the quarter, due to strong microprocessor sales growth and increased shipments of 90-nanometer units.

Incremental profit fall through on incremental CPG sales was 0 in the quarter, due to the high seasonal marketing and long-term brand investments.

For the first quarter of 2005 marketing, general, and administrative spending level is expected to return to third quarter levels.

Switching to the memory business.

Fourth quarter sales were $504 million down 11 percent from the fourth quarter of 2003 and 6 percent from the third quarter of this year.

Flash memory sales were weaker than expected, due to an aggressive pricing environment, particularly weak sales in Japan, and a delayed new product qualification in the wireless segment.

However, MirrorBit-based products increased as an overall percentage of the sales in the quarter, unit shipments increased, and ASP declined from the third quarter.

Disappointingly, the Memory Group lost money in the quarter, with an operating loss of $39 million, down from an operating income of $15 million in the prior quarter.

The Memory Group's gross margin declined in the quarter reflecting the aggressive pricing environment, only offset somewhat by record bit shipments.

Turning to the balance sheet.

As outlined at our November analyst day, we made significant progress in the fourth quarter delevering our balance sheet.

And as planned, we reduced our debt by $200 million in the quarter.

We exchanged for stock $201 million of convertible senior notes, reduced our near-term debt by retiring $612 million of our Dresden Fab 30, as it's secure, term loan, and issued a 7.75 percent senior note due 2012, thereby refinancing a near-term loan with a long-term one.

As stated earlier, these actions contributed to a $49 million, or 13 percent, charge to earnings in this quarter.

Cash balances ended the fourth quarter at $1.2 billion, flat from the third quarter.

Fourth quarter capital expenditures were $470 million, up as expected from $407 million in the third quarter.

Inventories increased from the third quarter by $68 million, primarily in leading-edge 110-nanometer Flash memory technology.

Now I'd like to summarize a few comments about the year 2004.

For the fiscal year 2004, we were at an all-time high for the Company at $5 billion, up a significant 42 percent from 2003.

Annual sales for CPG were $2.5 billion, up 29 percent from 2003.

Flash memory sales concluded 2004 at $2.3 billion, up 65 percent from 2003.

AMD's operating income for the year was $222 million, roughly a $455 million improvement.

In CPG we had an operating profit of $303 million for the year, and in Flash memory, we reported a profit of $35 million for the year.

This was good progress in both businesses.

Cash flow from AMD operation was positive for year at $1.1 billion.

For the year, EBITDA was $1.4 billion, a $602 million improvement from the prior year.

Capital expenditures for 2004 were $1.4 billion, under our guidance of $1.5 billion.

Now I'd like to discuss the outlook.

AMD's outlook statements are based on current expectations.

Following statements are forward-looking, and actual results may differ materially depending on market conditions.

Based on strong acceptance of AMD64 technology, AMD expects its processor amount to increase as the year unfolds.

In the first quarter, AMD expects processor sales to be flat to down slightly, in a typically seasonal down quarter.

AMD expects Flash memory sales to be down in the first quarter, due to the continued imbalance in supply and demand, continued pressure on ASPs, and seasonality.

For your modeling, consider the following -- In 2005 total research and development dollar expenses, and as a percent of sales, are expected to be up, primarily due to Fab 36 start-up costs of around $200 million.

First quarter R&D expenses are anticipated to be up approximately $20 million from the fourth quarter.

Total marketing, general, and administrative dollar expense are expected to be up in 2005, but as a percent of sales, are expected to decline.

First quarter marketing, general, and administrative expenses are anticipated to return to third quarter levels of 2004.

Our quarterly tax rate is anticipated to be within the 10 to 20 percent range throughout 2005.

In fiscal 2005, capital expenditures are expected to be around $1.5 billion. 2005 depreciation and amortization is expected to be approximately $1.2 billion.

Operating cash flow is anticipated to be around $1.2 billion, with free cash flow relatively neutral for the year, with the inclusion of Fab 36 grants and subsidies.

In summary we have made significant progress over the last couple of years to position AMD as a viable and financially sound competitor in the microprocessor and Flash memory markets.

With record 2004 and fourth quarter sales, an improved balance sheet, and despite seasonal trends, we are confident that AMD is well-positioned as we head into 2005 to continue to impress our product and technology leadership on the microprocessor and Flash memory markets and to maximize shareholder value of our 2 principal businesses.

With that, I'll turn it back to Hector.

Thank you, Bob.

As we mentioned, 2004 was a record year for AMD in many ways, but the finish was disappointing.

We're taking bold actions to address these issues and more aggressively capitalize on the opportunities before us and to maximize shareholder value.

In that context, AMD is entering 2005 with a formidable and unrelenting sense of competitiveness.

A desire to demonstrate that our overall success in 2004 is only a point on the line towards reinventing the competitive balance of the industry.

We have many formidable assets -- a clear and demonstrable performance advantage across a growing set of product segments; an expanded portfolio of world-class partners, companies like HP, Lenovo, Sun Microsystems, IBM, and others who see AMD as a path to real differentiation and unprecedented customer satisfaction; an evolving community of top-tier enterprise customers, evangelists willing to share their successes with their IT peers; and a solid ecosystem of partners, vendors, governments and communities that are investing in our success; a design and technology team that is second to none; an expanding portfolio of leading brands from AMD Opteron to Athlon 64 and 64 FX, AMD Sempron, and the newly announced AMD Turion 64 for mobile technology; a stellar manufacturing record in some of the world's most challenging technologies; and technology leadership momentum, highlighted by our AMD64 and MirrorBit technologies, and underscored by our leadership in the move to multi-core processing; and an employee base of highly-skilled and motivated employees who recognize the moment-in-time opportunity that's before us.

Most important we have developed a taste for sustained success, both in terms of the work we do and the economic prosperity that it produces.

This success will continue.

We are leading the industry to pervasive 64-bit computing, and we will increase our lead, bolstered by our leadership in multi-core processors and the upcoming release of both the Microsoft 64-bit Windows OS, and the highly-anticipated Solaris 10 operating system.

We are reinventing the Flash memory business on the strength of our innovative MirrorBit technology, and our revolutionary ORNAND architecture.

And we are determined to make this business provide, one way or the other, a fair return to our shareholders.

And we are empowering a new global generation of tech-capable consumers with a recently-launched personal internet communicator, and our emphasis on serving the high-growth regions of the world such as India, Russia, Latin America and China.

Like 2004, we are determined to do great things in 2005, and to continue to pursue our destiny to change the competitive dynamics of our industry.

Thank you, and now I'd like to turn it back to Mike for questions and answers.

Okay.

Operator, if we could start the Q&A session, please.

Thank you.

Ladies and gentlemen, if you would like to ask a question, please depress star, then 1 on your touchtone phone.

You will hear a tone indicating you have placed in queue, and you may withdraw your request from this queue at any time by depressing the pound key.

Our first quarter is from the line of Mark Edelstone from Morgan Stanley.

Please go ahead.

Yes, good afternoon.

I had two questions, if I could.

Bob, it sounded like from your commentary that the gross margins in the processor business were actually up sequentially.

Can you give us a general range as to how much the increase, is that, like, a 100 to 200 basis point sequentially?

More than 1, less than 2, Mark, but did definitely up.

Okay.

Great.

And than just a clarification on your comments on the balance sheet, that inventories were up primarily in 110-nanometer Flash.

The distribution reserves were also up in the quarter, I think for the second quarter in a row.

And I think in the prior quarter that was due to a richer mix of processors.

Can you talk about that as well, and to what drove that increase here in Q4?

As we continue to ship more and more AMD 64, which has much higher value to us and to the whole marketplace, that number continues to increase.

Actually, units are not up, it's actually a dollar shift.

So you'll see that trend continue as time moves on, just from the shift -- I will call it -- to higher-valued products.

Okay.

But that did not have an impact on your own balance sheet, on your own inventories?

Memory -- microprocessor inventory was up a little bit, but we had a pretty good quarter.

We hit record units, so we actually shipped a lot of product.

But most of our inventory build, as I highlighted, was in the Flash memory area, where, as Hector said, we were disappointed, and we actually missed our sales expectation, and built some inventory.

But all the inventory we did build was the leading-edge 110-nanometer-type product, particularly for the wireless segment.

Okay.

Thanks a lot, guys.

Thank you.

Thank you.

And our next question is from the line of Adam Parker from Sanford Bernstein.

Please go ahead.

Hi, in the Flash memory business, given bit demand, it looks like it's strong, what did you underestimate about the NOR Flash pricing environment?

I think, Bob, you mentioned supply-demand is a problem, yet, it, kind of on the surface, looks like Intel's got capacity and others do, and you guys are investing a lot of CapEx in the Flash.

So, how do you think the capacity environment for Flash will shake out in the coming quarters here?

And what can you guys do over the next couple quarters to generate profits again in Flash?

Let me make some comments about the capacity and the technology.

And then I'll ask Henri to comment about the pricing environment in Flash.

First of all, the marketing total is unbalanced right now.

There is definitely a lot more capacity in place than there is demand.

It is my judgment, my view, and my opinion that there is across-the-board, in general, except for isolated cases, an underutilization of capacity today.

And but the one being used is competitive pressures are pretty high.

So we expect that that will continue even though bit demand is still pretty strong, and will continue to increase over time.

We -- well, our plans, frankly, based on the results of the last 2 quarters, are now being modified as we speak.

We intend to drive hard, fight hard, drive utilization and capacity better, and be more competitive, more aggressive in the marketplace.

We thought and made a judgment call that beginning late in the fourth quarter and early in the first quarter there would be a stabilization and more normalcy in the capacity-demand equation.

That has not happened.

And as a result we're going to modify our strategy and aggressively go after business.

Henri?

Hector, from an environment perspective, although overall, obviously, the quarter was very challenging with particular pressure in Asia and within Asia and Japan, where the imbedded market was a little softer than expected.

And, overall, if you look at our business around the world, we had a much better performance in the wireless segment than the imbedded segment, where the pressure was the most severe with some ASP declining over 30 percent quarter-to-quarter?

Okay.

Just as a follow up, I remember, I guess, in the fall of '02, I wrote a piece that maybe AMD should exit the Flash memory business and Ben and Jerry told me in the most polite way that that was not going to be possible.

I guess I'm wondering what would have to transpire here in NOR Flash for you guys to consider, sort of, exiting the business, or what your thoughts about that would be depending on the environment in '05?

Well, first of all, let me tell you that -- that the Flash technology is unique and pervasive.

And our invention of MirrorBit is also uniquely positioned.

I think that's an awful -- and one example of that is when you look at the fact I have $504 million revenue in the quarter, we lost $39 million, which I hate, it makes me puke to lose $39 million.

But in all -- when I -- when you think about it, put it in perspective, I don't think there is a Flash company in the NOR space anywhere in the world with that price, that cost competitiveness that will allow you to lose only $39 million on $504 million in sales.

So we -- and we're a long way from, yet, milking the benefits of MirrorBit.

Having said all that, we are prepared to do whatever it takes to make sure that we get value to our shareholders going forward.

Okay.

So you're leaving that open-ended then?

We are open to anything that will allow us to increase value to our shareholders?

Okay.

Sorry, one last quick housekeeping question.

What do you guys expect depreciation to be in 2006?

So I'm trying to understand how you have flat depreciation 2 years in a row here when your CapEx is ramped so significantly from '03 through '05?

This is Bob.

When -- in particular I'll call it in last year, the investments we're making in Fab 36, which is the disproportional amount of money the Company is spending last year and this year, that does not depreciate until we actually turn the facility on to manufacture product.

Right.

So it's -- it's not in the depreciation rate.

So depreciation will step up fairly significantly in the '06 period of time as we start turning on the billion-plus investments we're going to make in Fab 36.

Fab 36 over the 2-year period was roughly $600 million last year and an equivalent $600 million this year of investments that will turn on to depreciation in the beginning of '06.

So, like, another 100 million a quarter from this year's rate or?

Divided by 5 years, to give you a ballpark.

Divide the 1.6 billion by 5 years.

Okay.

Thanks, guys.

Our next question is from the line of Ben Lynch from Deutsche Bank.

Please go ahead.

Yes, hi, guys.

Just a couple of questions.

First one -- is there any signs of Intel becoming more aggressive in the server space, and has that had any sort of impact on the business, the processor business in Q4?

Again, I'll make a few comments and Henri can follow up.

You know, when somebody asks me that question, I have to smile because we are in a very competitive space, and I don't know of a segment where AMD and Intel are not aggressively competing big time.

So to me that question is sort of like, yeah, it is tough, it is competitive, and do we see anything unusual, just doggone, you know, fight hard in the marketplace?

No.

Henri, would you like to add to that?

No, just underline that, Ben, we've had, again, a record revenue for server in the quarter, so we got continued momentum.

We were expecting our competitor to aggressively seed their new Nocona technology in the marketplace.

We've seen that.

But, frankly, on some of the key parameters that matter for the customer like thermals and efficiency, that technology is not revolutionary, and the leadership that Opteron providers to the customer remains intact.

And we, in fact, expect that as we move into 2005 and we introduce our dual-core based Opteron technology, we'll further enhance our competitive positioning in that market.

So, yes, it's tough out there, as Hector outlines it, but customers put a premium on the value of architecture and that's not changing.

Okay.

Great.

And then the second question I had was, Flash revenues and operating profit were down a lot in what's typically the seasonally better quarter.

And now for Q1, you're guiding that down because of continued price pressure and seasonality.

Without getting -- not trying to force you in a corner on this -- you probably have at least better visibility than us there.

Is there sort of sequential change or environment for Flash now?

Is it comparable to what it was in Q4?

Is it worse, is it better?

Just help us think about that, please?

Well, first, what we've got to do in the first quarter is accelerate in Japan as we outlined.

We have a challenge there.

If you recall then when we put together 2 organization, historically, AMD has a very strong presence in the wireless base.

Our partner, Fujitsu, had a strong presence in the embedded space.

In Japan, the wireless market a booming market.

We need to do a better job at penetrating that market, but we don't expect that to turn around within the first quarter.

So that -- the area of focus for us is to go and aggressively defend our position in the wireless space, which is still very strong, with all of the key customers around the world, and to be more aggressive and attack the embedded opportunities that are in front of us.

So the intense -- so the intensity of the issues facing you is going to be just as hard in Q1 as it was in Q4, I guess.

From a mark -- yes, the answer is yes, the intensity is going to be there.

And we are going to -- as I mentioned earlier without being too specific right now -- we are going to take some aggressive action to contain cost, accelerate technology, because one of the good things that's going to come out of this is that some of the underutilized capacity we have in place and some of the energy in our people is not going to be totally focused on figuring out how we can even move faster on technology and new products and take advantage of these market condition that exist.

Because the cost competitiveness of MirrorBit is pretty awesome, and the fact that we can get there totally, the better off we're going to be.

So we are going to redirect energy and resources to accelerate that rather rapidly.

Great.

Thank you.

Our next question is from the line of Michael Masdea from Credit Suisse First Boston.

Please go ahead.

Yes.

The first question, and I have a quick follow up after, is on the strategy overall.

Obviously, you guys have done a great in coming out with processors that compete -- that are very competitive with your competition, and that's had a good impact.

But your competition is talking a lot more about platform strategies than they have in the past, restructuring around that.

Could you just talk about how you guys compete against that and how important it is to compete against that?

We believe that the market -- the customer really would like to see all of us more focused on total solutions, whether you call it a platform or not.

And I believe that doing that makes sense, makes sense for us, makes sense for the customer.

And as a matter of fact, although I rarely do that, I'll pay a compliment to my competitor.

I think that's very innovative to do and it is almost tempting to ape it.

But I'd like to make a comment on the difference on how we might go about it, is that I think that the opportunity for us to establish deeper and stronger relationships with the ecosystem to provide customers the best solution, not the solution that we dream up in the lab and come up with and then cram it down their throat, but to be able to provide the best possible solution, and the combination of support infrastructure to accomplish that.

We believe that we are a company capable of doing that.

And our strategy, going forward, is to really strengthen that.

And you're going to see some moves in 2005 that are going to clearly indicate a strong move in the direction of being a strong ecosystem provider in total platform solutions, although in a different approach than our competition.

Just a follow up, ASPs are weaker this quarter.

You talked about that coming from the new -- the high-growth areas, presumably emerging markets or other areas.

Can you just talk about business was, and kind of on a like-for-like, apples-for-apples, other parts of the world where, some of maybe the lower-growth areas?

Sure.

I want to compliment Hector's answer to your first question.

As you may have not noticed, but when we announced our new Turion 64 mobile technology, it is not only a processor technology, but also, it really is a platform offering, that over time with partners, we will present to the market.

Now turning to your second question, what you need to understand in the mix of the quarter is that we actually had stable ASPs in our server business, it actually grew a little bit, the same in our Athlon 64 business, and in our mobile Athlon 64 business.

What we've seen through the quarter is a very high demand for our new Sempron brand in high-growth markets, and we decided to serve that demand.

So the ASP variation is really a mechanical effect of the higher mix of Sempron product than we expected initially.

Thanks a lot.

We have a question from the line of Joseph Osha from Merrill Lynch.

Please go ahead.

Hi, can you hear me?

Yes.

Hector, you appear to be picking up Jerry's pension for colorful rhetoric.

Two questions, first on the microprocessor side, can you update me as to what percentage of starts or outs, whatever you prefer, is 90-nanometer, and just give me an update as to what their road map is there?

And then I have a follow up.

Right now we're -- and I -- within some first-order approximation, we're pretty close to over half of our starts being 90-nanometers.

And our plan is to be, for all practical purposes, converted totally by the second quarter.

The reason I say practical is because there are some products that have unique customer requirements that may keep us for some time longer at 130, but that's just strictly just to serve a customer need for not changing the platform.

So, by the end of the quarter, we'll be -- the second quarter we'll be fairly much done on 90-nanometers.

And there is no way, I am going to agree that could ever could dream to be as colorful as Sanders is.

You have your own charms.

Second question, Bob, I admit I'm struggling modeling operating expenses here.

I've been looking for SG&A to be up by about the same dollar amount as it was in the third quarter, and I was surprised there and also surprised on R&D.

So forgive me for being literal, but can you please, again, nail down for me what you expect in dollar terms for both line items in the first quarter, and then to the extent that you can, dollar terms for both line items for all of 2005?

Sure, Joe.

First on the R&D line, for the first quarter, we will continue to see that increase from the fourth quarter levels to the tune of about $20 million from the fourth quarter level, so increasing by $20 million as we move in.

That is driven by Fab 36 start-up costs as we continue to hire more people and continue to exercise the equipment.

That line, in total, by the end of the year will be up $200 million year-on-year.

And then at that --

It will ramp as we go quarter-to-quarter.

Right.

And then in '06 it, presumably, starts being absorbed through cost of sales as you ramp Dresden, right?

Absolutely.

You got it.

Okay.

In the SG&A, marketing, and general and administrative area, first quarter was a -- I mean, fourth quarter was a high watermark.

First quarter will drop to third quarter levels.

So that delta we picked up quarter-on-quarter of about $42 million will go away.

And for the year, I would model marketing, general and administrative expenses will be up in absolute dollars, but as a percent of sales, depending on what you pick, it will be down.

So we'll see some dilution as we go through the year.

But first quarter, we will quickly move back to third quarter levels.

Okay.

And then just a last quickie follow-up on the inventory numbers.

Did you flow through any reserves relating to Flash memory pricing at all, or Flash memory costing more properly?

No, actually by building more inventory of -- in the Flash business we actually took on more reserves.

So we actually continue to be conservative in valuation in the memory business, particularly in this pricing environment.

I'm sorry, explain that?

You build more reserve?

So gross inventory, just pick a number, gross inventor went up by 100, and then inventory went up by 85.

All right.

So, you flowed through some reserve value.

You flowed some reserves, I'm sorry, you're realizing reserves?

That's right.

All right.

Thank you, very much.

Thank you.

And our next question is from the line of Tim Luke from Lehman Brothers.

Please go ahead.

Thank you.

I was wondering if you could provide some color on how you see the gross margin developing going forward?

Thank you.

In both businesses?

Well, let me do it in both businesses.

Thanks.

First in the microprocessor business, because clearly the Company is dependent on the mix of the business, because the gross margins between both businesses are significantly different.

In the processor business, we're actually, as we said, we're north of 50 percent.

We improved quarter-on-quarter.

And as we continue to penetrate deeper in the commercial space, both with Opteron and client, we will see further expansion in the gross margin, and further shipments of 90-nanometer.

So we feel like we will continue to see improvements in gross margin in the microprocessor business on the continuum as we go through time.

In the Flash business, it is very pricing dependent.

It's hard to anticipate right now improving gross margin in the first quarter with sales looking like a down quarter.

Our -- so right now, we are modeling, we have budgeted the memory business to improve gross margin year-on-year, and that is predicated mainly on continued execution of making MirrorBit a significant portion of what have we shift in that portfolio, which clearly has a cost advantage, which you clearly need in a very aggressive pricing environment.

Could you remind us what those metrics are for MirrorBit as a percentage of the mix, then, as we move forward?

The current, we're -- we're in the double -- we're in between 10 and 20 percent of our business today is MirrorBit, and by the end of the year, we expect it to be 50 percent.

That's right.

Just one last clarification on the Flash side, were the units up sequentially?

Yes.

Units were up.

ASPs were down, which, obviously, down more than the units.

And both would have been down sequentially at the beginning of the year?

You know, seasonally it's hard to tell at this point.

Right now, seasonality would say units would be down.

ASPs continue to be drifting in that direction.

Thank you, very much.

Thank you.

And our next question is from the line of Krishna Shankar from JMP Securities.

Please go ahead.

Yes, in the Flash business have you resolved the product trade that you talked about in the wireless segment?

Yes, we had a 256-Megabit product that we had hoped to qualify in the fourth quarter.

We had a glitch that required us to change a number of layers toward the end of the process, because there toward the end of the process we could address them rather fast.

And as we speak, there are some customers that are already qualified the product, and we have a number of other customers in the process of doing so.

So we expect that quarter -- that product to have an impact on sales this quarter.

Okay.

And on the processor business, can you elaborate a little bit more on the Turion 64 introduction and the platform strategy that you talked about earlier, both in terms of timing and features for the Turion 64?

For the Turion 64 mobile technology was officially announced at CES.

We intend to have partners shipping the technology in the market towards the second part of the second quarter.

And as far as the platform strategy, we have a series of ecosystem partners that are going to join us and provide Turion 64 certified platform that include, you know, both chipsets and wireless solutions to form a comprehensive and cost-optimized solution for the mobile market.

And the final question is on how AMD sees the adoption of PCI Xpress and DDR2 memory, both for the PC space and server space, going forward, and your performance product lines?

All right.

So, on the PCI Xpress, obviously, there is an aggressive conversion that's being driven in the industry.

We have a number of platforms available on PCI Xpress.

And we believe that that conversion will happen very rapidly on the basis of the cost structure of the solution.

When it comes to DDR2, we believe that that transition will be slower because today DDR1 solution are still cheaper.

And in particular, in the case of the AMD architecture, there is no performance benefit to moving to DDR2.

So we obviously have a strategy to move to DDR2 technology at the right time, and we're working on that timing with our key partners.

Thank you.

And our next question is from the line of Andrew Root from Goldman Sachs.

Please go ahead.

Thanks, very much.

I'm curious as to what kind of traction you can report seeing in the enterprise desktop space following on your success in server, if that's started to radiate out at all into the desktop, and if that has any impact on the first quarter outlook, which is a little bit more robust and seasonal?

Well, obviously, when a Fortune 500 company is entrusting AMD with its server choices, which is a strategic choice for the enterprise, that bodes well for our ability to go and present client-based solution.

In fact, out of the 40-plus members of the Forbes 100 list of companies, some of them already use both Opteron and client-based AMD solutions.

I'm not going to disclose the details of this, but, obviously, or us as we walk into 2005, further penetration of the commercial client segment is a key priority on the strength of our Opteron offering.

How many of the -- would any of those 40 companies using your client solutions have been customers before, or those are all new customers (inaudible)?

They're all new customers.

Okay.

And then, how much, is the Sun launch of your new Opteron server line materially impacting your Q1, or is that sort of, we will see (inaudible)?

One of the key differences between Q4 and Q1 when it comes to Opteron will be the number of platforms available to the market.

If you look at our Opteron offering, there were still certain segments that were not available.

For example, competitive blade solutions, more of the commercial-type of implementation of the Opteron technology.

As we outlined in our analyst conference, all of these segments will be covered by our OEM partners in the first quarter.

And when it comes specifically to Sun, one of the key influence that Sun has helped us tremendously in the ecosystem is in the software area.

They are driving a very aggressive expansion of 64-bit based solutions in the software ecosystem, of course, in preparation of the highly anticipated launch of the Solaris 10 operating system.

And just one last quick question on Flash.

Is there a price decline in the first quarter that would dictate a more significant breakdown of existing inventory than what it will cost to market?

We don't think so at this point.

We've been pretty conservative of our inventory evaluation.

So the answer would be no, we don't think so.

So, could you absorb first -- fourth quarter-type move again in the (inaudible)?

Absolutely.

Okay.

Great.

Thank you.

And our next question is from the line of Tom Thornhill from UBS.

Please go ahead.

Yes.

Could you review again the cash flow character -- parameters, and particularly relative to what was laid out in the November analyst meeting?

Yeah.

Cash flow from operations, I won't go historically, you can pick that up.

But going on a forward basis is capital expenditures will be $1.5 billion, which is identical to what we talked about at the analyst day.

I'm flipping through my notes to make sure I say the exact right numbers.

Depreciation for '05, you should model about $1.2 billion.

And cash flow from operations will be about $1.2 billion, with an assumed increase in working capital, because we are assuming the business will continue to grow.

Free cash flow will be relatively neutral, but that is counting the grants and subsidies we're going to receive from the European governments for Fab 36.

Okay.

And then looking at the operating margin on CPG which came down a bit in the December quarter, but with MG&A coming down and gross margin looks like it's holding, should we -- should operating margins come back to the September quarter levels?

Or better.

Or a little bit better?

Yes.

That's right.

All right.

As we talked about, we made fairly significant marketing investments in the second half of '04 to provide some liftoff, as we called it, for AMD64 client-based products.

Most of those programs are coming to an end.

So you will see a drop back to more historical rates in the microprocessor area in marketing.

So with continued good gross margin, a little less investment in those areas will yield better operating income.

Thank you.

Thank you.

And our next question is from the line of Kevin Rottinghaus from Midwest Research.

Please go ahead.

Yes.

Could you remind us of where you are with 110-nanometer transition in Flash, and if there's a -- what's left to generate as far as cost savings out of that?

The 110-nanometer conversion to Flash will be fairly done here in the next quarter, completely.

And now we're aggressively, as I mentioned earlier, due to the fact that we see a slow down in the market for at least a couple of quarters, we're going to take advantage of that to do everything we can to accelerate the next-generation even more than we had already planned.

So although we're pleased with the fact that we led in the conversion to 110-nanometers, that's probably going to be a short-lived technology and move rapidly to the next node, which will be somewhere around 90-nanometers.

As far as the cost that -- benefits of that, we're experiencing those as we speak.

And one of the reasons I think we did not experience heavier losses in the quarter is that the cost competitiveness of that technology is fairly solid.

We expect that to continue and improve throughout the rest of the year.

Okay.

The transition or the qualification issue you had, was that at all related to either MirrorBit or to 110, or was that more just a manufacturing issue?

No, it was one of those things that happens in this complex technology.

First of all, it was a MirrorBit part.

And it was a layout issue with part of the die, where when you layout this very complex high-dense parts, just a tiny, little gate somewhere in the wrong place, and we had to redo the mass head and do it over.

Okay.

Thank you.

And our next question is from the line of Glen Yeung from Smith Barney.

Please go ahead.

Thanks.

Hector, you gave earlier in the call, some expectations for what MirrorBit would be as a percent of mix, which I think were the same numbers that you had given out in the third quarter.

And I wanted to get a sense, listening to you discuss accelerating some technologies in Flash, whether or not we should expect that mix to maybe accelerate a bit faster?

I think that would be the hope and the plan, and we're going to work as hard as [expletive] to make that happen.

It's too early to try to change the guidance that we gave you before.

What are the actual barriers that prevent from you actually doing that any faster?

Well, there are a number of things that affect that.

One is customer qualifications.

You know, customers that are today in a floating gate device that to sign in their works, have to not requalify and justify the cost of that change.

We think we -- the attraction that allows us to induce for to change is the fact that the cost is significantly different and customers are responding.

But at this point in time, it will be hard to forecast if we can actually change that rate of response at this point.

Okay.

And then just one other thought here, which is, at the analyst day, you talked about your outlook for 2005 PC unit growth of 10 percent.

And now that we've come another month and change since that meeting, I wanted to get your sense as to whether or not you think 10 percent is still a reasonable target for '05, and how you think you're going to gain share in that year, i.e., is it fairly linear throughout the year that you'll keep gaining share, are there any sort of milestones we should be looking for?

I don't think there's anything we've seen in the last few weeks since the analyst that changes our view of 2005.

We're very optimistic about 2005 in the IT world.

I still believe that 64-bit technology will make an impact in 2005, as I said in the analyst meeting.

I think we have an upcoming number of events that are very powerful, like the highly anticipated Solaris 10 release.

It's an operating system, as well as Windows 64, which will happen in the first half of this year.

All of that, I think, and combined with the value equation for our products makes us very optimistic that at the end of 2005, we'll be looking like we did today, where in 2004 we grew 29 percent in the processor business, which is significantly better than the market.

And we're certainly counting that we will also do that again in 2005.

Very good, thanks.

And our next question is from the line of John Lau from Banc of America Securities.

Please go ahead.

Okay.

Thank you.

Back over to the Flash.

You mentioned, Henri, that the ASPs will be drifting down -- drifting downward.

For our modeling and longer-term contracts, can we expect additional pricing impact as new contracts such as handset Flash start this quarter?

No.

I think, as you know, John, we have a pretty good visibility.

The wireless market is concentrated on relatively few accounts, with whom we have long-term relationships.

So this is not where surprises are expected.

The embedded marketplace, which is far more fragmented and relies heavily on the channel is where we've seen a lot of the turmoil.

And, frankly, a lot of the smaller players in the NOR Flash market are taking desperate action to just manage for cash.

Great.

And then, finally, what is the typical seasonality that you anticipate for Flash in Q1, if you were to take a look at the overall market?

Your question is from a revenue perspective, units, bits?

Actually, let's say, for example, units, typical for Q1.

I believe flat, to down a few percent, maybe between 0 and 5 percent down.

Great.

Thank you, very much.

And our next question is from the line of Michael McConnell from Pacific Crest Securities.

Please go ahead.

Thank you.

Question on unit growth we've seen now in the industry.

So if my model is correct, you guys saw about 11 percent unit growth in Q3.

Looks like your units were well north of 15 percent in Q4.

If you look at your competitor, they had units up about 15 percent in Q4 as well.

Could you talk a little bit about inventory out there in the marketplace?

Are there any, what kind of revisions you're putting out there to be sure there's no double ordering going on with some shortages out there in the channel, just anything on that side would be very helpful?

Sure, first, our channel business had a record quarter, but it's been the case throughout 2004, both in units and in dollars.

We continue to have to drive a very aggressive conversion of what we we call a tray processor, which are typically OEM processors, to what which we call product-in-a-box, which is a retail version of our processor.

That, again, hit an all-time high as a percentage of the total channel business.

And as weeks of inventory, we exited the year at the same level that we entered the year.

And we have a policy to keep our weeks of inventory between 4 and 5 weeks.

So we keep a very close leash on that aspect of the business.

You're relatively comfortable that double ordering, at this point, is not -- you're not having any of those issues with customers?

No, absolutely -- absolutely not.

In fact, at the very high end of the spectrum, in some weeks, demand exceeds supply.

It's a very strong market.

Okay.

And then another question I have was just on Sempron, looking at, kind of, what happened with Q4, it was a surprise to some people.

Is this looked at as -- could you talk a little bit, kind of, what the mix shift, what went on there in Q4?

I assume most of it was a mix shift to Asia.

Is this something that's going to be just a 1-quarter issue?

Or is -- do you expect Sempron sales to continue to exceed that of AMD64 and lay a mix as we kind of get through the first -- the next 2 quarters or so of 2005?

Thank you.

Well, I think, what we've seen in Q4 is really the result of a good positioning of our AMD Sempron brand.

You're correct that part of the demand came out of Asia, but, frankly, the high-growth markets of Europe were also showing very strong demand.

We continue to see also very good success in Latin America.

And it seems to me that in those high-growth markets, the Sempron brand is very well accepted and, perhaps to a degree, is demonstrating better value for the user than our competitor.

So, I don't expect that demand to go away.

We, obviously want to make sure that in those markets we have a proper mix, and we participate in all aspects of the business.

And we don't want to saturate the market with the ultra-value parts.

So it's a matter of managing the brand and managing price positioning of our solution.

Thank you.

And we have a question from the line of John Barton from Wachovia Securities.

Please go ahead.

Yes, thank you.

On the topic of the Flash JV, just looking at the numbers of the operating loss, and then what was shipped off to Fujitsu, it would imply if I'm doing the math right, that the non-operating expense allocated to that division was about $3.5 million, which was down significantly from low-to-mid 20s over the last several quarters.

Can you give us some insight as to why that number is changing the way it is, please?

I'm not sure I follow your math.

I mean, we continue to have shared service models between the 2 business -- between AMD and the business.

That has not changed very dramatically.

Of course, we've tried to rein in as much cost as possible with the weak business environment.

We will continue to do that on a go-forward basis.

The operating loss reported was $39 million.

It was about 16, 17 million shipped off to Fujitsu, which, like, represents 40 percent of the overall loss.

And so the number I'm backing into is the difference between the operating loss and the total loss.

And that spread looking like it's about 3.5 million versus, again, something in the mid 20s.

Am I missing something in the bath -- in the math, Bob?

No, I think you got the math right.

So I -- I don't know that specific answer to your question, I guess is the real issue.

But we can -- I think off-line we can figure that one out together.

Fair enough.

Then on the topic of gross margins, again, just trying to triangulate back into your data, my model would imply that gross margins for Flash went from somewhere in the mid 20s to, call it, somewhere in the mid-teens quarter-on-quarter.

Am I in the ballpark on those numbers?

You're not wildly crazy.

Okay.

Then, finally, the $49 million charge for the debt restructuring, from a line-item perspective, all the interest in other or was that broken across multiple lines?

No all -- except for the 2.9 million that's sitting in the restructuring line, the rest of it's in the interest income and other line on the P&L.

Okay.

And I lied.

One last one.

Just from the convertible interest add-back perspective, relatively 0 in this quarter, I believe, and then what are we look at going forward?

0 for this quarter because we lost money.

So you're not seeing the diluted shares come in from the convertible.

So it kind of depends on your model of how profitable you want us to be.

But clearly, it'll be back in -- like it was in prior quarters, all of the other quarters of 2004.

But at lesser count, remember, because we did retire 200 million of the 400 million 4.5 percent security.

Thank you.

Our next question is from the line of Chris Danely from J.P. Morgan.

Please go ahead.

Thanks, guys.

Bob, just a follow-up on John's question.

So that means that you guys think you'll be profitable in Q1?

Yes.

Great.

And then, just another net on the CPU gross margins.

So you think that they go up or stay flat in Q1 with the revenues of flat-to-down?

We think we're on a path to continue to see progression in gross margin.

It's obviously easier in a growth environment of the top-line than in a shrinking environment.

I mean, there's no surprise in that.

But with our cost reductions, and as Henri says, the continued mix management to actually participate in every segment of the market, we believe ASPs can continue to move in a northward basis, as they did, really, throughout 2004, except for the fourth quarter.

And, therefore, gross margins will improve.

Okay.

Great.

And last question, on the inventories, it's getting up above 100 days.

What are your plans for utilization rates in the first half of the year?

Do you plan on bringing down that inventory and bringing down utilization rates, or just keep running like you are right now?

It kind of depends on business.

In the microprocessor business, we run every wafer we can run, maximizing the revenue for that facility based upon where we are and where we need to be for our customers.

Flash, to be honest, we actually slowed down a little bit in the fourth quarter.

It was a little late.

But we're actually reviewing that as we speak to actually be more aggressive, as Hector said, to accelerate technology migration total 90-nanometer node, increase MirrorBit as a percentage of the total and actually move this inventory that we have.

Sure.

So on that point, it doesn't sound like the CapEx is going to change?

No, not at this point.

We'll obviously continue to review it.

As we have, I think we've demonstrated good financial scrutiny on CapEx to try to modulate that with the business environment.

Got it.

Okay.

Thanks a lot, guys.

Thank you.

And our next question is from Paul Leming from Soleil Securities.

Please go ahead.

My question has been answered.

Thanks, very much.

Operator, we're going to take 2 more questions, please.

Very good.

And our next question is from Allan Mishan from CIBC World Markets.

Please go ahead.

Hi.

Can you tell if desktop grew faster than server in unit terms this quarter, please?

Yes, we had -- are you talking -- in pure units, we had a slightly higher growth.

Great.

And being that Sempron, I guess, is a consumer-oriented product and Q1 is generally not the big consumer quarter, would you expect for your ASPs to trend back up in Q1?

As a result of the mechanical mix, yes, we expect ASP to trends back up in Q1.

Also because we expect, as I indicated previously, an acceleration in the Opteron server sales in Q1.

Right.

Okay.

Great.

And just one last housekeeping one.

Do you have a share-count number if you did the, I guess, what is it, $.05 without the $.13 charge?

It's right in the 400 -- yes, the number would be right in the $410 million -- 410 million share range.

Okay.

Great.

And just remind us what the break points are for when the different converts hit now that you did that change?

9 -- I think it's $.09 on the $200 million convert that's still out there, and $.20-some on the 500 million 4.75.

Mike will check that if you want to call Mike back, but I -- that's in a ball -- that's a pretty good ballpark.

It's 8 or 9, and just maybe north of 20.

Okay.

Thanks, very much.

Sure.

And our final question is a follow-up from Ben Lynch of Deutsche Bank.

Please go ahead.

Yes.

Hi.

Henri, it's a probably a question for you.

You guys have pretty consistently been gaining revenue market share from Intel the last 18 months or so.

And it looks like you lost some in Q4, although your unit share stayed sort of flattish.

When do you think we could expect to see you guys, just based momentum and stuff, resuming your revenue share wins?

Well, Ben, first I would remind you that, of course, we started from a pretty good performance in Q3 and had a high watermark.

So quarter-to-quarter, indeed, we're not pleased with the fact that we lost a bit of revenue share.

But on a year-to-year basis, we still achieved a pretty good performance in Q4.

As we indicated, we believe that in Q1, the acceleration of our commercial segment efforts will pay dividends.

And, of course, it's our strategy to continue to focus on gaining both revenue share and unit share, and not simply unit share.

One other comment I'd make, Ben, just make sure we're all -- we do not have a 14-week first quarter. 2005 is not our year to pick up that extra week versus our competitor.

So just want to make sure that's on the table, you don't assume that.

We are not the same.

We've only got a 13-week quarter.

Yes.

That's important as well.

Thanks, very much.

Okay.

Great.

That concludes the call.

Thank you, everyone, for your participation.

Thank you.

And, ladies and gentlemen, once again as a reminder this conference is available for replay after 7:45 p.m.

Pacific time today through January 28 at midnight.

You may access the replay service by dialing 1-800-475-6701 and enter access code of 765351.

International participants may dial 320-365-3844, and, again, use the same access code, 765351.

That does conclude your conference for today.

Thank you for your participation and for using AT&T Executive Teleconference.

You may now disconnect.