英特爾 (INTC) 2004 Q3 法說會逐字稿

Good afternoon.

My name a Derrick and I will be your conference facilitator today.

At this time I would like to welcome everyone to the Intel Q3 '04 conference call.

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

After the speaker's remarks, there will be a question-and-answer period.

If you would like to ask a question during this time, simply press star, followed by the number one on your telephone keypad.

If you would like to withdraw your question, press star, followed by the number two.

Thank you.

Mr. Doug Lusk, Director of Investor Relations, you may begin your conference.

Thank you and welcome to the Intel third quarter earnings conference call.

Attending are our CFO, Andy Bryant, and President and COO, Paul Otellini.

I would like to remind everyone that the earnings release and this call are available on our IR website at intc.com.

For those of who you did not see the earnings release, revenue in the third quarter was $8.5 billion, and earnings per share were 30 cents.

The third quarter earnings report discusses Intel's business outlook and contains forward-looking statements.

These particular forward-looking statements and all other statements that may be made on this call that are not historical facts are subject to a number of risks and uncertainties and actual results may differ materially.

Please refer to our press release for more information on the risk factors that could cause actual results to differ.

The specific forward-looking statements cover expectations for product mix and demand, revenue, gross margin, expenses, tax rate, interest and other income, capital spending, depreciation and amortization of acquisition-related intangibles and costs.

These statements do not reflect the potential impact of any mergers, acquisitions, divestitures or other business combinations that may be completed after October 11th, 2004.

Lastly, if, during this call, we use any non-GAAP financial measure, as defined by the SEC and Reg.

G, you will find on our website, intc.com, the required reconciliation to the most directly comparable GAAP financial measure.

I now want to introduce Andy Bryant, who will discuss the third quarter earnings results.

Andy?

Thanks, Doug.

The third quarter overall revenue growth, consistent with our forecast in September; however, revenue mix was somewhat different than we expected as shipments of microprocessors were notably less than we expected, while motherboards and chipsets were well above our forecast.

As one would expect, this mix change contributed to a disappointing decline in gross margin percentage.

Most of the major businesses grew during the quarter, led by mobile computing, chips sets, servers, flash memory, and wireless connectivity.

While revenue came in as expected, gross margin was at the bottom of the range due to the revenue mix, and also inventory reserves and valuation.

Unforecasted tax items not in our September update, that is approximately 3 cents to earnings per share.

The outlook for the full year points an annual growth in the double digits for revenue and gross margin dollars, with gross margin percentage that is approximately flat to slightly up.

The ability of the business to generate cash has been consistently high.

Returning some of this cash to stockholders, we increased the stock repurchases for the quarter by $1 billion to 2.5 billion total.

Revenue of $8.5 billion was up 5% from the second quarter, at the lower end of the pattern typical for this period.

Compared to the third quarter of 2003, revenue was up 8%.

Unit shipments were higher than the second quarter for Intel architecture microprocessors, chipsets, and motherboard.

In Intel communications, unit shipments of flash memory were approximately flat, while those for wireless connectivity products were higher.

Revenue in the Intel architecture business was $7.1 billion, up 5% from the second quarter and 4% from a year ago.

Within this group, the sequential dollar growth was split nearly evenly between microprocessors and the combined revenue of the remaining product.

Microprocessor revenue of $5.9 billion grew 3% sequentially, and 3.7% year-to-year.

The remaining revenue up $1.2 billion, which comes mostly from chipsets and motherboards, was up 18% sequentially and 6% year-to-year.

Revenue in the Intel communications group was $1.3 billion, up 4% from the second quarter and 36% higher than a year ago.

Nearly all of the sequential growth came from flash memory and wireless connectivity products.

Gross margin dollars of approximately 4.7 billion were essentially flat with the second quarter and up 4% from the third quarter of last year.

On a percentage basis, gross margin was 55.7%, just below our revised September forecast of 58%, plus or minus a couple of points.

The difference was, in part, a function of revenue mix.

Demand for microprocessors was lower than we anticipated in September, while revenue from lower margin products, such as chipsets, motherboards and communications products, was higher.

In addition, wafer costs were lower than we anticipated for the chipset known by the code name "Grant Sales," resulting in an inventory revaluation.

And finally, microprocessor inventory reserves were higher.

In comparison to the second quarter, gross margin percentage is down by about 3.5 points.

The primary reason for the decline were higher unit costs of microprocessors, inventory reserves across all major product lines, and the shift in revenue mix to lower margin products.

In comparison to results in the third quarter of 2003, gross margin percentage declined 2.5 points.

For the company, overall, operating income of $2.4 billion was flat with the second quarter, and up 3% from the third quarter a year ago.

In the Intel architecture business, operating income of $2.8 billion was 39% of its revenue.

Operating income is flat, sequentially, and down 4% year-to-year.

The communications businesses took a step backwards on the progress toward profitability, as losses doubled from the second quarter.

The flash business achieved good growth in revenue, but flash units in some segments were lower than we originally anticipated in July, and inventory reserves were consequently higher.

In addition, our flash businesses experienced higher inventory write-downs as a result of lower unit costs.

Although these developments could make it more challenging for the communications businesses to achieve our goal of profitability in 2005, profits remain the goal.

We are currently working through our an annual planning process with this in mind.

Spending on R&D, marketing and G&A was $2.3 billion, lower than our September forecast and down slightly from the second quarter.

As a percent of revenue, spending was down slightly from the second quarter and the third quarter of 2003, and it is also down slightly for the year-to-date.

The number of employees rose 3% during the quarter to 84,200 at the end of the September.

This represents year-to-year growth of 6%.

The total in the third quarter for interest income, other income, and gains and losses on an equity investment was $53 million.

In this category, interest and other income was $63 million.

Tax items increased earnings per share for the quarter by approximately 3 cents from our September update.

After preparing the 2003 tax returns, we reduced the third quarter tax revision by $195 million.

This lowered the effective tax rate from 29.5% forecast in September, to 21.4%.

The difference from our update was largely due to the fact that the expert sales deductions were higher than we expected, as were the state tax benefits related to last year's divesture efforts.

Diluted earnings per share, which includes potential dilution attributable to employee stock options, was 30 cents.

Basic earnings per share, which does not include potential dilution, was also 30 cents.

Average shares for calculating diluted earnings per share was 6.4 billion.

Basic shares outstanding were also 6.4 billion, down 1% from the second quarter, and down from the peak in the second quarter of 1998 by 6%.

During the quarter, we repurchased 106 million shares at a cost of $2.5 billion, an increase of $1 billion over the previous two quarters.

The objective of the buy-back program is to return cash and increase the return to stockholders.

On the balance sheet, accounts receivable were up $83 million, or 3% from the second quarter.

Day sales outstanding were 35, down from 36 in the second quarter.

Overall inventories were lower than than the second quarter by $43 million, or 1%.

The decrease in working process was much larger, lower by is $162 million, or 8%.

These changes reflect some of the same dynamics behind gross margin.

Inventory reserves adjustments were higher than we expected.

These changes reduced inventory levels, but also lowered the gross margin percentage.

As we said in July and September, we have been working to lower the inventory levels of microprocessors by slowing the planned growth rate of wafer start of these products.

Inventories are still higher than we would like, and we will continue the effort to reduce them this quarter.

Cash, short-term investments and fixed-income trading assets ended the quarter at $15.8 billion, down from the second quarter by about $800 million.

After the stock repurchases -- after the stock repurchases -- somebody is coming through. [Technical Interference.] Excuse me, operator --.

Cash, short-term investments and fixed income trading assets ended the quarter at $15.8 billion, down from the second quarter by about $800 million.

After the stock repurchases of $2.5 billion, capital spending of 1.1 billion, and dividend payments of 253 million.

As we turn now to the outlook for the fourth quarter, please keep in mind that the forecast data do not include the effect of any new acquisitions or divestitures that may be completed after October 11th.

I will use the midpoint of forecast ranges when making comparisons to prior periods.

We expect revenue in the fourth quarter to be between $8.6 and $9.2 billion.

The new point of this range would be sequential growth of 5%.

We should have the lower end of patterns typical for the fourth quarter.

In the five years prior to this, growth in the fourth quarter has ranged from flat to up 12%, and averaged 8%.

The forecast is tempered by the assumption that customers will continue to work off inventory levels in the fourth quarter.

It also anticipates modest pricing pressure in microprocessors and less sequential growth in revenue for flash memory.

Compared to the fourth quarter of 2003, the midpoint of the forecast revenue range would represent growth of 2%.

Gross margin percentage in the fourth quarter should be approximately 56%, plus or minus a couple of points.

This would be flat with the third quarter, and lower than would be typical with the projected revenue growth.

By slowing the rate of growth away from starts to reduce inventories, we will also increase unit costs in the near term.

At the same time, the modest pricing pressure I mentioned will probably lower product margin.

We also anticipate charges for underutilization of 200-millimeter capacity and startup costs of 65 nanometer process technology.

Spending, R&D plus MG&A, should be between $2.4 and $2.5 billion, an increase from the third quarter spending of $2.3 billion.

Depreciation should be between $1.1 and $1.2 billion.

Amortization of acquisition rate intangibles and costs should be approximately $40 million.

Our estimate for gains and losses from equity investment and interest and other income was a net gain of $65 million.

The tax rate for the fourth quarter is expected to be approximately 30.5%.

For the full year, other aspects of our outlook are largely unchanged.

These include capital spending of $3.8 billion, plus or minus $200 million, R&D of $4.7 billion, slightly below the previous forecast of $4.8 billion, depreciation of approximately $4.6 billion, and amortization of acquisition-related intangibles of $180 million.

The nine-month results, combined with the our fourth quarter outlook ranges, point to a year with revenue in a range of around $33 to $34 billion; gross margin percentage around 57 to 58%, and gross margin dollars of roughly $19 billion.

This translate s into revenue growth for the full year of 10 to 12%; gross margin percentage is approximately flat to slightly up from 2003, and an increase in gross margin dollars for the year of approximately to $2 to $2.5 billion, or 11 to 15%.

With that, let me turn it over to Paul for an additional comment on the business.

Thanks, Andy.

In the third quarter, our Intel architecture business saw growth at the low end of seasonality, while the communications group had a slightly higher revenue, driven by flash and wireless land connections.

A highlight for the quarter was record unit shipments in both mobile and server microprocessors.

We also reached a key technology milestone, as the majority of CPUs shipped in the quarter were built using our 90 nanometer production process.

On a geographic basis, our Asia-Pacific region continues to set new records and broke the $4 billion mark for the first time.

Revenue there was up 10%, sequentially, and 23% versus a year ago, led by good demand for chipsets, motherboards, and cellular flash.

We set revenue records in Taiwan and South Asia, which includes India .

China continues to experience double digit, year-over-year growth and set a revenue record for a third quarter.

In [aMia,] revenues were up 13%, sequentially, and 12% versus last year.

We saw good demand in Western Europe in both consumer and corporate, especially for mobile products.

Emerging market revenue within aMia was slightly higher, but tempered somewhat by lower than expected sales in Russia.

In the Americas region, revenue was down 8%, sequentially, due to an inventory correction at some of our large, OEM customers, along with a continuing shift of PC production to Asia.

In the end markets, corporate demand was solid, led by strength in mobile and server, while U.S. retail sales were disappointing.

In Latin America, we had solid growth as the economic picture there continues to brighten.

In Japan, sales were approximately flat with the second quarter, with strength in notebooks, exports, offsetting weakness in the retail sector.

Our channel business was up slightly in the quarter and set new revenue and unit records for a third quarter, led by growth in emerging markets.

Moving to our Intel architecture businesses, the mobile group had another strong quarter with record units and revenue.

Our 90 nanometer processor, code-named ["Doe San,"] ramped quickly and represented well over 50% of Pentium M processors shipped.

Demand for Centrino mobile technology continues to be strong in corporate, and gain momentum with consumers in the back-to-school selling season.

Our enterprise group had record units in the third quarter and posted revenue growth of over 20% versus last year.

During the quarter, we introduced our new Xeon platform for dual processor servers that provide support for faster DDR2 memory, high bandwidth PCI express interconnect technology, demand-based switching for power reduction, and 64-bit technology for large data sets.

We are very pleased with the market acceptance of this new platform and expect it to quickly become our volume server product.

At the Intel Developer Forum in September, we demonstrated a forthcoming Itanium family processor code-named "Montecito," that will have 1.7 billion transistors providing dual course CPUs and 24 megabytes of cache memory.

Our desktop business experienced weaker-than-expected demand in large part due to inventory corrections at a couple of our large OEM customers.

Our 90 nanometer Prescott chip volume grew significantly during the quarter, more than doubling, and crossing over with Northwood.

Our chipset business gained market segment share again in the third quarter, and had record units led by the ramp of our [Gransdale] family of products, which began shipping in June.

Although the early momentum of Gransdale was stalled by a few weeks due to a manufacturing excursion, we expect to exit this year with Gransdale representing approximately half of our performance desktop shipments.

Our motherboard business also had a strong quarter with sequential unit growth of over 40%, setting a new record.

Our communications group grew revenue 4%, sequentially, and 36% versus last year.

Flash revenue increased by 9%, led by good demand for our higher density cellular products, allowing us to regain the No. 1 spot in the newer market segment.

Our wireless LAN products for the Centrino platform had another strong quarter with an increase in attach rate with Pentium M processors at a greater than 30% sequential growth in units.

In cellular, Intel, Nokia and Symbian announced plans that will make it easier for manufacturers to build phones based upon the Intel silicon, the Nokia Series 60 platform, and Symbian OS.

In addition, Intel and Symbian are developing a reference design that manufacturers can use to build 3G phones, based upon Intel's processor, code-named "Hermon."

At IDF, we announced that we have begun sampling a system on a chip code-named "Rosedale," for [YMAX]-based broadband equipment for homes and businesses.

In summary, although growth in the quarter was not as high as we expected in July, we had a number of bright spots, including record unit CPU shipments for mobile and server, as well as making substantial progress on the desktop platform with the ramp of 90 nanometer products and the Gransdale chipset.

We also believe we gained market segment share in flash, chipsets, motherboards, and wireless LAN connections.

Going forward, we plan to use our leadership in process technology, architectural innovation and platform capabilities to drive growth in our IA and communication businesses.

With that, let me turn the meeting back over to Doug.

Okay.

Thanks, Paul.

We will know open the call for Q&A.

We will attempt to take questions from as many participants as possible.

To help, we ask that you please limit yourselves to only one question and no more than one brief follow up.

Thank you.

Operator?

[Operator Instructions.] We'll pause for just a moment to compile the Q&A roster.

Your first question comes from Tom Thornhill with UBS.

Thank you.

Looking at the fact that you're underloading the fabs at this point to absorb inventory, at -- how far below standard loadings are we and how long do you think that's going to have to run at this level to get inventory back to target levels?

I give a different answer, Tom, between 300 and 200 millimeters.

Let me do those separately.

Three hundred millimeter are not fully loaded, but they're above what I would call the minimum standard water mark for our no forecasted excess -- underutilization charges on the 300 millimeter network.

And if fact, everything I can see says that we're at the low point for those factories now.

As far when we are, quote, at the levels I would like to be on products coming out of those 300 millimeter factories, if demand holds as we see it right now, for the fourth and first quarters, we'll be okay on those particular products, sometime towards the end of this year.

The 200 millimeter factories are a little bit different.

They are -- we are taking underloaded charges in the fourth quarter on those.

And in our case, that's -- I don't want to get too specific as to where we draw the mark -- but we will have notable underload charges in the fourth quarter.

We will start to reload those factories in the first quarter;

I suspect it will take until the second quarter before I get those back where I want to be.

So consider the older products, I've got one to two quarter's worth of work to do on the leading edge products.

This quarter, I should be okay by the end.

And a follow-up on, again, back on the standard cost issue.

Previously, the 300 millimeter dividend, if you will, was expected to reduce unit costs, I think, from -- I think the slide you had used previously was Q3 -- Q4 '03 to Q4 '05, by about 25%.

Can we tell that you're achieving those standard costs so that when we return to optimal loading, or standard loading, will we still achieve that dividend, if you will?

Sure.

If we can achieve what we would call "normal" loading on the 300 millimeters factories, we'll make that call.

So what it really is going to come down is through next year, if we get those factories back into full production, I'm not worried about the cost target.

If we don't, then we'll have a problem.

And you should be able -- you were saying that the inventory should be cleared on -- at the 300 millimeter product level by the end of -- you said this year;

I assume you meant '04.

Yes, by the end of '04.

So we should get back toward standard loadings in Q2 at least, or the benefit of it by Q2?

Assuming we're having "normal demand" next year, yes.

So it's going to come down to the strength of the market next year.

If we have a decent market, we'll be at full loadings.

We'll be fine.

If we have a weak market, we'll be unloaded.

Thank you.

Your next question comes from [Eric Gomberg] with Thomas Weisel Partners.

Hi.

Good afternoon.

I was hoping that you could maybe discuss what you're seeing for underlying demand trends in Q4.

It sounds like your processor business might be a little bit less than seasonal, particularly on the desktop and retail side.

If you could separate the excess inventory that your customers may be holding, what do you think the amount actually looks like?

I think Andy characterized it.

We've pegged the midpoint of the revenue at what would be the lower end of our -- what we've seen -- average seasonality standpoint.

But the range on average seasonality, as he pointed out, is really quite broad, given the kinds of years we've had in the recent past.

I think that we would have been looking at a much more normal seasonal target for revenue had we not had the residual inventory impact of our customers.

Okay.

Then just skipping back over to inventory on your balance sheet, could you perhaps give some quantification of what the reserve was in dollars in Q3 and where you think inventory could go in Q4 if you were to hit the midpoint of the revenue range in the midpoint of margin guidance?

I won't answer your question specifically;

I'll try to give you a little bit better feel for what we're talking about.

The simple question was the second one.

I don't want to put a number on the fourth quarter, but if I -- my goal is to get inventory down more in the fourth quarter than I got in the third quarter by a reasonable amount.

So you can get a sense for where we think we're headed.

In terms of the reserves, there were -- versus the September update -- there were three major factors that caused margins to be lower than expected.

One was the product mix between -- we'll say microprocessors, the higher margin products, than the other products.

One was what we call "inventory revaluation."

If you think of what inventory revaluation means, I'll give a specific: On the Gransdale chipset, the wafer costs are lower, which means I now have the benefit of -- in the future -- of a lower cost per unit.

But all the inventory that's in the line has to be valued at that new lower cost.

So I take a revaluation, or "an inventory write-down" for that.

The third factor, our reserves, and then the reserves come in two different buckets.

There's one that I'll call a good bucket of reserves.

So we're building -- the risk starts on a future product called a Prescott 2-meg.

If that product qualifies, it means when we start to ship, we have more available to get the ramp started quicker.

If that product doesn't qualify, it means the extra product I'm starting and building now, I basically have to throw away.

So that's why we call it, it's a risk start.

We did more of that than we expected, which means we're better prepared, assuming the product qualifies when it's supposed to.

So that's one of those reserves I'd look at and say, that's a good thing we're getting more product out of the process.

It's optimistic, assuming the product works.

Another one would be something like [old Prescott steppings.] We have some older Prescott steppings, which we held in die, and were planning on building through to become Celeron products.

Based on the fact that the yields of the new production are better, I actually can save cash by building new parts and getting a higher yield to the end product than by shipping the old one.

So I write those off.

An then there's some basic quality reserves, and those operate in the negative category.

So it's a lot of words we're saying.

It's not as simple as a single reserve.

Of the three categories I gave you, one is mix, one is revaluation, and one is inventory reserves.

They all are an enough of a size to be noticed.

The inventory reserves are slightly bigger than the other two.

Next question, please.

Your next question comes from Adam Parker with Sanford Bernstein.

Yeah.

Andy, think you lost me there with the reserves.

Okay.

So --

So let me start [inaudible] again.

Of the three buckets, reserves are the biggest of the three.

But all three are big enough that they would be noticed.

Okay?

Step one.

Okay.

Okay.

Of the reserves, there are three different things.

I'll call some are okay reserves and some are bad reserves.

When we start Prescott 2-meg product, because we don't yet have a qualified to ship for revenue, I don't value that material.

So it's called -- it's reserved.

And my gross inventory, I calculate how much of it would have been worth and then I reserve it.

So when we talk about reserves, that's one way you get a reserve.

Another way you get a reserve is you take an old Prescott, basically an early on -- early build product from last year, which could have been made into product and say, from a cash point of view, it's inefficient to build that through, so we're going to throw those away and we reserve them.

The third one would be if I have a yield problem or I'm not getting as high quality as I want [inaudible] and that's -- that means I'm throwing away more of the die I'm building, and that's what I call a bad reserve because it's a quality reserve.

So what was the total magnitude of the reserve?

Again, I didn't give you the specific.

I said I -- I said versus update, which I expected to be approximately 58, plus or minus a couple and I'm actually at just under 56.

So I'm two points down.

Right.

I have three elements, of which all are notable, and the reserve is the largest of the three.

That's all the granular I'm giving.

All right.

So if I -- so my question is: that wasn't a question.

Okay. [Laughter.]

I didn't understand.

So what's the real question?

My question is just trying to understand your margin guidance, which encompasses understanding this inventory reserve thing.

But basically, you have flat depreciation, sequentially.

You've got 5% growth.

And I'm trying to figure out, this is one issue, this inventory issue; the second is, can you talk about the magnitude of what's in your forecast in terms of the underabsorption charge of 200 millimeter, and what's in your forecast in terms of 65 nanometer startup costs that hit cogs?

Okay.

So again, I'll give you generalities and let you, kind of, pinpoint your own triangles.

If you look, going into the fourth quarter, with -- give you the midpoint of revenue with $400 to $500 million of revenue growth --

Right.

-- you expect margins to go up.

They're not going to go up.

The biggest drivers of them not going up -- there are actually three, again -- the mix is one, but the bigger drivers are the underutilization charges on the 200 millimeter network and the start-up costs on the 65 nanometer network.

Of those two factors, the underutilized, underload charge, is the bigger of the two factors.

Okay.

Now -- sorry, one follow up -- is on the 65 nanometer start up, why don't we see R&D go down?

In other words, isn't this shifting engineers to cogs from R&D like you did last year when you finished 90 nanometer startup and sent them to 65 nanometer process development?

Some of that, but it's also inside a current factory that's running.

And what you'll tend to see is cost flowing into the running and the starting of the new process, as opposed to being into a product cost someplace else.

So one of the reasons your inventory does go down is you direct more effort into these spaces.

But what about R&D?

Does that go down?

Not much.

You know, again, you'll see some timing and some change, but it's not much.

Okay.

I'd ask more but I'm sure you don't want me to.

Bye.

Thanks.

Your next question comes from [Michael Mazdaya ]with Credit Suisse First Boston.

Yeah, I guess a couple of questions on the demand side.

Is there any kind of correlation that you guys have seen out there between what you see in the second half and the first half?

And kind of within that, you made a comment about first quarter and what you were expecting.

If you could just elaborate on that, that would be great.

If I commented on first quarter, then I misspoke.

All I said was, if we had "what we expect."

And we're not giving any expectations yet for the first quarter.

I'll let Paul talk about second half versus first half.

Well, he asked if there was any correlation between what we're seeing in the second half and with a we saw in the first half.

And I guess in hindsight, the answer has got to be: yeah.

We had a more robust view of the third quarter and the second half three months ago, when we made the lost last conference call.

That was based upon first half results and what we thought our second half demand was from our customers.

As we went through the quarter, and we and our customers understood their inventory better, particularly some large OEMs, we were able to -- we brought our numbers down.

And that's what you're seeing from us now in terms of our, kind of, integrated view of the year.

I guess what I was trying to get at more is given we've seen a little bit lower end of seasonality in the second half, you know, it's more consumer driven.

That's coming in the lower end of seasonality.

Does that typically mean that you see a weaker enterprise in the first half, or have you ever seen that kind of correlation as you look forward?

No.

No.

The corporate demand is generally good worldwide.

I think if we were to go back and replay the year and been perfectly intelligent about inventory downstream of us, I think you would have seen a more linear year and a more normal shape of the year.

A more normal shape.

It's tough to go back and [inaudible--microphone inaccessible] inventories.

What Paul is saying is we think Q2 is a little higher than seasonality and that's exactly what's probably on, and it makes the rest of the year unfold pretty normally --

Yeah.

-- maybe a little weakness in U.S. retail.

Right.

And when you were talking about the manufacturing outlook for next year in utilization, you mentioned that a normal year.

Is a normal year kind of a seasonal year and is that the key kind of watermark, or what would you consider a normal year?

Again, I don't want to get into defining a "normal year" and what normal growth in this industry is.

If you look at the industry growth over the last few years and say you have something approaching that next year, it should be enough to allow to us keep our factories pretty full.

Great.

Thanks.

Your next question comes from John Lau with Banc of America Securities.

Yes.

Thank you.

Your guidance for next quarter is very conservative.

I was wondering, in what specific sector do you see issues or concerns that brings your expectations down, and is it due to more macro economic factors and is there any visibility into what the back -- I guess the holiday season brings?

Thank you.

Well, in reality, as Paul said before, when we look at the fourth quarter, what we see is still some inventory, we believe, in the hands of our customers that needs to be worked off.

We could see a little bit of weakness in U.S. retail, but that's about the only place.

So in reality, we're not saying a big sea change out there; we're seeing a catch-up to inventory problem that began in the second quarter with a little bit of demand effect, but not a big one.

Okay.

So a lot of that is some of the inventory left over then?

That's we believe.

We believe there's a bigger effect in Q3, but there's still some left.

And when you say inventory, in this case, we're talking inventory held by our customers not held by Intel.

Nor the channel.

Or the channel.

Or the channel, thank you.

Your next question comes from Apjit Walia.

Thanks.

First question is just about the desktops.

Do you believe the softness you're seeing is more of a cyclical quarterly event, or maybe it's a bigger trend, a secular or structural problem.

And what do you plan to -- how do you plan to address that?

Well, the desktop thing you really have to pars between mature an emerging markets.

And in emerging markets, the first PC, and most PCs, are still desktop based.

And as I reported, most of our sales in Asia-Pacific, which had a record, and our channel sales, which were at an all-time best third quarter for Intel, are reflective, I think, of the emerging market desktop trend being rather robust still.

In mature markets, there's a couple of trends.

One is the overarching trend for corporations to repopulate desktops with notebooks, and I think that's one of the reasons you're seeing mobile sales hit records.

And then the other is, where are we in the replacement cycle.

I think on top of that, the one, as Andy said, the one soft spot we did see in consumer sales was in U.S. retail.

Okay.

And sort of a big picture question on the semi cycle.

I mean, you're talking about inventories being primarily a reason for some softness.

And say that you took the other argument in the cycle, say peaked in the middle of the year, and now we are accelerating and continue accelerating into '05, if you look at most macro indicators.

Now, your margins, typically, track the cycle very well, so that means margins will come down significantly as we go into '05, if you look at historical correlations.

And if you look at that, then how do you -- if that is the case, then how do you address this?

I mean, aren't you going cut CapEx and take some more steps to address this, if this is actually not an inventory issue but a pure in-demand issue?

The -- I'm not ready to make a capital forecast for next year, so I'm going to probably frustrate you in your question.

I'm also not ready to declare next year a down cycle.

At this point, what I would say is our goal is to try to make sure we get our new, modern modern factories full.

If it turns out we have excess capacity, you deal with it in the old 200 millimeter network, not in the 300 millimeter network.

You continue to do what we did in the last turn, downturn, which is bring out products that hopefully give you increased differentiation versus your competition, and wait for the next cycle to come up.

This is not a timely retreat.

Great.

Thank you.

Your next question comes from Andrew Root with Goldman Sachs.

Thanks.

Can I ask you a question on your cash and your uses of cash?

Cash was down about 815 million in the quarter.

I mean, not material relative to your overall balance, but you bought back more stock, obviously, than cash generated.

If you could just sort of refresh us on your thinking about cash uses, strategically, going forward, especially in the light of the comments you just made, Andy, about not wanting to declare a downturn but, obviously, things not as robust as maybe you were thinking about six months ago.

And just talk about how you expect to be deploying cash and balances, et cetera, over the next couple of years?

You know, it's difficult to give a definitive answer to the question.

In general, we do [inaudible] cash down about a billion dollars.

We increased the buyback in an effort to return cash to the shareholders.

We, overall, think that's a good thing.

I am not able to give you a forecast for buy backs in the coming quarters.

I'm not able to actually answer much of the question.

In general, yes, we have a lot of cash.

Some of it needs to go back to the shareholders.

We did some of that in the third quarter.

Okay.

But I guess, suffice to say, if your business continues to be strong, there's no reason it doesn't seem like strategically you feel like there's any need to build up cash at this point?

I don't believe there's a strategic reason to build up cash at this point.

Okay.

Maybe an obvious statement.

But, and then Paul, I just want to go back to one thing that you mentioned, just so I can understand in terms of your customer behavior, if you could just give an indication if the destocking is expected to be more serious on the desktop or notebook side, and then enterprise and consumer and any assessment of what might cause that behavior to turn around?

Well, it's hard to tell.

I think it's going to be predominantly on the desktop side.

You can't differentiate between consumer and corporate processors.

Okay.

I think that the indicator I gave you on Gransdale and the ramp, in terms of its percent of our premium desktop products, is a very good indicator that both corporate demand and uptick on that product line and our view of corporate markets.

Okay.

Fair enough.

Thank you.

Your next question comes from Chris Banely with the J. P. Morgan Chase.

Great.

Actually, have I one quick clarification, Andy.

So it sounds like utilization rates go down a little bit in Q4, and do they bottom in Q4 and then stay flattish in Q1 or move up in Q1, assuming normal seasonality?

Again, it will be different by the different networks.

Actually [inaudible--microphone inaccessible.]

But overall, overall for the whole company?

If you -- well, again, because there's such different product lines, I hate to give you an answer to that -- a cavalier answer.

The underload charges in the fourth quarter are the 200 millimeter network.

All things being equal, I would expect to still have some in the first quarter, not as much.

And I have no underload charges on the 300 millimeter network, and I would expect, if anything, to more fully utilize those assets in the first half of next year.

Okay.

And then, I guess, just one quick follow up.

Obviously, there's some inventory issues out there.

This would be a question for you or Paul.

I'm just wondering what you guys -- or how you guys view your competitive positioning out there, and do you think that you perhaps could have lost a little bit of share to your main competitor?

Well, given what I told you about record numbers in mobile and server, I think our competitive position is really quite good.

Given that most of the inventory that we are correcting ourselves from, and our customers, is desktop.

When we went back and looked at the -- what we believe the impact of that inventory bubble was -- it would have given us -- and adjusted our numbers for that over the last three quarters -- it would have given us, essentially, flat market segment share Q1, 2 and 3.

If you allow for that inventory bubble to -- which happened in Q3 -- to reflect in Q3, only it would have taken us down slightly.

So basically you're saying you think you might have lost a little bit of share in Q3?

As a result of the bubble.

And I don't know if you guys can answer this, but how do you think that happened or what would you do to address that, and then I'll stop there.

Let me help Paul.

What he's trying to say is in terms of PCs purchased in the marketplace, shares stayed the same in quarters one, two and three.

The only changes were in inventory levels between us and our customers.

And their work in process of raw materials.

But in terms of shipments out, we think share between Intel and its competitors were the same in each quarter.

Okay.

Thanks.

Your next question comes from Glen Young with Smith Barney.

Thanks.

Maybe just as a follow up to that last question.

If we were to maybe zone in on share and emerging markets, for example, and there seems to be some evidence that your competitor maybe gained some share in China, for example, is that something that you see in your business and if that is, is there something specific in that market that you may be targeting?

I don't believe that we lost share in Asia.

There was one notable change in one major customer in China, but China is a lot more than one customer.

It's principally a channel-based business and our record sales in Asia and our all-time best third quarter in the channel ever, I think reflect our continued strength in that market.

Okay.

Fair enough.

And then the other question, I just wanted to talk a little bit about more of the server market in terms of the pull you're seeing now for [Nakona] and [Lindenhurst] and what the demand levels are, specifically, for those devices, and the strength that you're seeing?

We had record server units this quarter, up 20% year-over-year.

A lot of that is triggered by the new shipments on Nakona and Lindenhurst.

We've had great acceptance.

Virtually every major every customer has launched multiple SKUs with it, and it's ramping.

Great.

Thanks.

Your next question comes from Tim Luke with Lehman Brothers.

Thanks.

Andy, just to be clear with respect to your inventory comment, you said that you thought the fourth quarter inventory level would be down by a margin that was slightly more than the third quarter decline, where I think you said 43 million lower.

Is that what you meant or it going to be significantly down from that level?

And within that, in addition to the reserve, it looks like there was a write-down as a result of lower chipset unit costs.

Could you give us a sense of the magnitude of that write-down and how that might have played into the down 43?

The -- what I tried to say about inventory in the fourth quarter is I expect it to be down by more than inventory was down in the third quarter.

Inventory in the third quarter was down 43.

I don't think I said the slightly word, but if it's just slightly it won't [inaudible--microphone inaccessible] my expectation.

So down notably more in the fourth quarter versus the third quarter than the third quarter versus the second quarter.

As far as quantifying exactly how much the chipset evaluation is, I really don't want to get into that level.

Specifically, I didn't even quantify the entire reserve amount.

What I would tell you is if you look inside the detail in the inventory, microprocessor inventories were down more than 100% of the total drop in the third quarter.

Okay.

And just for Paul, you talked about flash.

And should we assume that you expect to continue gaining share in the calendar fourth quarter and how do you see the pricing developing in that arena going forward?

Thank you.

Well, it's certainly our intention to gain share.

We've had that program all year.

As you know, we're up this last quarter, in terms of revenue, pricing was okay.

The fourth quarter will play out depends on the what the mix of OEM versus broad market and flash ends up being over the course of the quarter.

But our goal is to being competitive in both segments.

Thanks.

Your next question comes from Joseph Osha with Merrill Lynch.

Hi, Andy.

I'm going to join the ranks of the confused here.

You said at the end of the second quarter that half or so of the inventory issue related to the 90 nanometer ramp and that working out maybe a little better than you expected.

And then you just said that microprocessor inventories, which I presume or at least in part, significant part, 300 millimeter 90 nanometer were down.

How did does that square with the fact that your 300 millimeter loading appears to be okay and isn't going down.

I can't put those pieces together, and then I have a follow up.

Okay.

So you made some leaps there that I don't think we gave.

We didn't say --

Okay.

-- loading was down again.

We said loading was less than full.

We also said we were ramping the --

No, you said inventory was down.

You just indicated that more than 100% of the inventory dropped, that you just recorded, was microprocessor.

Yes, that's true.

If you go back and look at the inventory detail, the total's down $43 million.

If you look go look at work-in-process, which is where you see the effect of slowing the ramp on wafer starts, it was down $150 million.

So the effect of the slower -- slow wafer starts is absolutely having an effect.

You see it in the work-in-process.

And that's mostly a microprocessor statement.

It actually isn't a big surprise, as you're ramping Gransdale into -- a new park into it's build, as your growing share in motherboards, which is a material-intensive environment.

You're tending to add material in those places while you're reducing inventory in the CPU space.

Okay.

Just -- but, again, the question was you were characterizing at least half of your inventory problem in the first half of the year as being related to microprocessor, 90 nanometer, but now you're characterizing your underutilization is driven more by your older, 200 millimeter network and I just can't get that.

Actually, if you remember in the second quarter when we were surprised by inventory build, I said half of it was in the flash and other products.

And half was microprocessor.

You said almost the half of the net additions to inventory was microprocessor.

Correct.

And now what we're doing is wer are reducing the microprocessor piece and adding some of the other pieces.

But you were able to do that without taking any underutilization charges to your 300 millimeter network.

Correct.

We did not take underutilization charges.

We did have lower utilization than we originally had planned.

Okay.

And that relates to my second question.

You've talked about there being two elements to the margin weakness.

One is the inventory, the actual valuation charges that you're taking.

Which, by the way, one simple number would be nice, if you've got it.

To what extent -- how long is it till we start to see -- assuming things work out seasonally as you expect in your 300 millimeter network -- the benefits of higher utilization manifest themselves?

Are you hoping that you can maybe get the 300 millimeter network working back to where you'd like it at the end of Q1, say?

So again, no question's an easy question, right?

Yeah.

If you have a 13-week throughput, which is approximately what we have in these processes, it says your unit costs are more reflective of your prior quarter's build than the current quarter's build.

Right.

So in my fourth quarter, I still don't have those factories working at what I call full level.

I don't necessarily in the first quarter, either, but I have them "fuller."

So what I think you will still see is you'll see costs -- higher cost per units in the fourth quarter this year, based upon getting the inventory down.

You'll see that in the first quarter and you'll start to see recovery in the second quarter, assuming demand is " normal," whatever you define normal to be.

So with our expectation demand, I would expect to see still relatively high costs in Q1, and start making progress again after that.

Okay.

You've been kind enough if you look at back, say, at the memory translator hub issues and stuff like, to throw a number out there.

Can't we just get one overall inventory hit number out of you guys?

Not this time.

If it were one big item, I would.

But again, I was serious, when I go through and look at these inventory values, I've got a series of 5 to 7, 8 to $22 million things.

So it's very difficult to call them out because there's not a single thing that's important enough to be called out.

Okay.

Thank you very much.

As a group, it kind of matters.

Okay.

Thank you very much.

Your next question comes from David Wu with Wedbush Mortgage Securities.

Yes, hi.

Two questions, please.

The first one is, you folks mentioned about a lower ASP in the fourth quarter this year.

If your consumer business is supposedly lower, what would lead to a lower ASP from Q3 to Q4?

The other one I have, actually, is regarding the mix of business.

In the old days at least when you have good chipset shipments and motherboard shipments, you tend to have microprocessor following.

Is the formula working, or is this time around it doesn't work?

Let me get your second one first.

I think it does work.

In general, particularly in the third quarter, David, chipsets and motherboards lead PC sales and CPU insertion.

I think that this third quarter we saw, as I said, records in chipsets and motherboards.

The increase in those numbers, quarter-to-quarter, I think is more than a leading indicator to processors.

In both businesses, we increased our market segment share, in our opinion.

Okay?

So that increment in market segment share isn't necessarily a precursor to processor increments.

Uh-huh.

Okay?

On the ASP, I don't think we said, though, that consumer would be lower in Q4.

I think we said that U.S. retail was soft in Q3.

Right.

In the fourth quarter, generally speaking, consumer sales are above Q3 consumer sales.

Right.

That's built into our model.

In general, I think that as Andy indicated, we've got some pricing pressure built into the model and are showing that there is some ASP erosion from Q3 to Q4.

All of that, at the end of the day, is going to be predicated on what the final mix between desktop, mobile and servers is.

Uh-huh.

Yeah.

I was wondering whether there are special programs to increase the inventory depletion?

No.

No.

We firmly believe that you control inventories through controlling what comes into and out of the factory.

That's a much more cost effective way than by lowering price and having flatter sales.

Okay.

Thank you.

Your next question comes from [Graham Tanaka] with Tanaka Capital Management.

Yeah, hi.

I just was wondering if you could -- what your models, with Intel's models -- would suggest on an ideal operating rate, in terms of having a cushion relative to effective demand when you risk adjust and look at the penalties as well as the rewards? [Inaudible--microphone inaccessible] As far as --

Graham, I'm not sure I understand the question.

Can you --

Well, there was an underproduction and a loss of market share one or two years ago, and I think there's a sense on my part that you would rather err on in terms of overproducing, especially in the second quarter.

I'm wondering if your models suggest that you should overproduce relative to what you think model demand is going to be, and if you could share that with us?

We always try to make sure we have a little bit of cushion.

Again, in this this business, you lose a lot if you can't meet your customer's demand.

So, yes, our model will always call for a week, maybe two weeks of inventory, to give us a buffer against a surprise demand spike.

So here what happened is because of the nonseasonal changes in inventory and demand during the quarter this year, it was a lot more than this, certainly one or two weeks?

Yeah.

But remember, also, we indicated that in the nonprocessor portions of the business, we're also growing and taking share.

So that says that you if want to plan a buffer, you need to plan a buffer for chipsets and motherboards as well as microprocessors.

[Inaudible.]

You have to lift your inventory a little bit, versus prior cycle.

Thank you.

I just wanted to comment a little on [inaudible--microphone inaccessible] you know, and what kind of [legs] you're seeing.

What kind of which we're seeing?

We can't hear you.

Centrino and the kind of demand follow-through you're seeing there.

It's got a very fast start.

I'm sorry, if that's a Centrino question, yes, it's doing extremely well.

It's one of the reasons we have a record mobile quarter.

I talked about the WiFi chips growing substantially, quarter-over-quarter, and maintaining a very high attach rate, which is what defines the Centrino kit.

So that's continuing into, you think, in the fourth quarter and building your models?

Yes, it is.

Thank you.

Operator, we'll take two more questions, please.

Your next question comes from Mark Edelstone with Morgan Stanley.

Good afternoon, guys.

Andy, wonder if you could give us a general sense as to how you think WIP will change in Q4, or as you exit Q4, versus what the change looked like in Q3?

Boy.

This is Andy Bryant thinking analytically.

I actually didn't look at the forecast by development of inventory.

I would believe we would take it down some more.

I think based on the way we're starting wafers in the factories and what we're doing in the 200 millimeter factories, WIP should come down some more.

Finished goods will probably stay the same.

So my suspicion is that we'll see a reduction in WIP.

And so would that suggest then that the relative loadings of Q4 would be lower than the change that we saw in Q3 already?

Certainly in the 200 millimeter factories, yes, and the 300 millimeter factories, probably not.

Thanks.

And then just one final question on the 65 nanometer startup costs.

Can you just give us a sense as to how we should expect those costs to ramp in the first half of next year?

And I would guess they would probably be peaking out somewhere around in the middle part of the year?

We're actually at a pretty reasonable level of those in the fourth quarter, the numbers I gave you.

I might see them being up some in the first half each quarter but not substantially different than where we are right now.

Thanks, Andy.

Your final question comes from Charlie Glavin with Needham.

Sorry for ending this on another topic.

Taxes.

But Andy, can you go through a little more detail in terms of the range in tax expectations from mid-quarter?

And given that a lot of that seems to have been implied as far as the amount of export shipments, I'm not sure within that if it was related to the ECI tax benefits, but if you could give a little more clarity as far as the change that did occur and why it will not carry forward, because I assume that corporate tax bill within Congress really would not affect that.

Boy.

A lot of different things in there and a pretty complex question.

There were two reasons for the change of $195 million versus September.

The bigger of the two was actually deductibility according to federal statute of income earned overseas.

So it's not all overseas income qualifies.

You have to go through the process and the spreadsheet to actually figure out what works and what doesn't.

But we had underestimated the level of income earned overseas that was deductible for tax purposes.

Andy, is that income earned but produced within the U.S. and then exported and recognized externally?

It's -- the income is earned overseas.

Okay.

The wafers done here, yes, but you don't get the tax benefit on the wafer; you get it on your assembly and test, which is someplace else.

And does that correspond with Paul's comments abouts the OEMs and shipments to Asian manufacturing?

No, not really.

It has to do purely with our production, and it has to do with us trying to figure out where in the world the profit will be made and then which of that profit will qualify under the U.S. tax code.

The second piece is you remember we did some divestitures last year.

We had the savings in the federal estimate already.

We had done some, obviously, not deep enough work in the state return and thought we had no real tax benefit.

Then we found out, upon actually starting to prepare those returns, that we get a deduction in some states as well for the value -- the deduction on those divestitures.

So it's both of those.

Again, the deductibility of the foreign source income is the bigger of the two.

And Andy, just to clarify, the biggest reason why it doesn't carry over into fourth quarter is because of the September 15th corporate tax deadline?

Well, the return we just filed was -- is not for this year's, it's for last.

We think the line rate that we had for this year is correct.

If you look at the new law, which isn't signed yet -- I don't think they got signed today -- it actually doesn't exactly replicate this system we've had in place.

So I'm actually having to look through and see exactly what it means to us.

My guess is we won't have as much benefit from foreign source income as we have had under the prior law.

Got it.

Thanks, Andy.

Okay.

Thank you.

We'd like to thank everyone for listening to today's call.

A recorded playback of this call will be available at approximately 5:00 p.m.

Pacific time tonight.

Those interested should dial 1-800-642-1687 and reference pass code 163506.

Thank you.

That concludes this evening's Intel Q3 '04 earnings conference call.

You may now disconnect.