輝達 (NVDA) 2008 Q4 法說會逐字稿

(technical difficulty) fourth fiscal quarter ended January 27, 2008.

On the call today for NVIDIA are Jen-Hsun Huang, NVIDIA's President and Chief Executive Officer, and Marv Burkett, NVIDIA's Chief Financial Officer.

Before we begin today's call, I would like to take care of some general administrative items.

Your lines have been placed on a listen-only mode until the question-and-answer segment of today's call.

During this call, we will discuss some non-GAAP financial measures about net income, diluted net income per share and gross margin when talking about our results.

You can find a reconciliation of these non-GAAP financial measures to the GAAP financial measures in our financial release, which is posted on the Investor Relations page of our website at www.nvidia.com.

Unless otherwise noted, all references to research market and market share numbers throughout the call come from Mercury Research or John Peddie Research.

This call is being recorded.

If you have any objection, you may disconnect at this time.

Please be aware that if you decide to ask a question, it will be included in both our live transmission as well as any future use of the recording.

Also, shareholders can listen to a live webcast of today's call via the Investor Relations page of our website.

The webcast will be available for replay until the Company's conference call to discuss its financial results for its first fiscal quarter 2009.

During the course of this conference call, we may make forward-looking statements based on current expectations.

Forward-looking statements, including the statements as to the importance of and the uses for the GPU; our outlook; the impact, performance and availability of and demand for our products and technologies; our growth and strategies; and the future of computing pertain to future events and are subject to a number of significant risks and uncertainties.

The Company's actual results may differ materially from results discussed in any forward-looking statements.

For a complete discussion of factors that could affect the Company's future financial results and business, please refer to the Company's Form 10-Q for the period ended October 28, 2007, and the Reports on Form 8-K filed with the Securities and Exchange Commission.

All forward-looking statements are made as of the date hereof, based on information available to us today.

And except as required by law, the Company assumes no obligation to update any such statements.

The content of the webcast contains time-sensitive information that is accurate only as of February 13, 2008.

Consistent with the requirements under Regulation FD, we will be providing public guidance directly in the conference call and will be unable to provide significantly more information in offline conversations or during the quarter.

Therefore, questions around our financial expectations should be asked during this call.

At the end of our remarks, there will be time for your questions.

In order to allow more people to ask questions, please limit yourself to one question.

After our response, we will allow one follow-up question.

I will now hand the call over to Jen-Hsun.

Thanks, Mike.

Good afternoon and thank you for joining us.

We are pleased to report another record quarter, and fiscal 2008 was another record year.

We had many achievements this past year.

Q4 revenue grew 37% year over year to a record $1.2 billion.

Annual revenue increased 34% year over year to a record $4.1 billion.

Annual net income increased 78% year over year to a record $797.6 million.

Annual gross margin reached a new high of 45.6%, an increase of 320 basis points from a year ago.

We launched a record number of industry-defining products and initiatives this year.

GeForce 8800 family, including the highly acclaimed 8800GT -- we've sold over 2 million 8800GTs in just four months of production, and demand continues to exceed our forecasts.

GeForce 7000, the first motherboard GeForce for Intel systems -- together with the CPU, the GeForce 7000 enables the first two-chip Intel processor PC.

CUDA, the industry's first parallel programming environment that puts the massively parallel processing capability of the GPU in the hands of all programmers -- CUDA is available on every one of the 50 million GeForce 8000 series GPUs that we have shipped.

We expect to ship several hundred million CUDA GPUs over the next several years.

CUDA has quickly become one of the most pervasive and accessible parallel programming environments ever.

There is increasing recognition that the traditional CPU-alone architecture or homogeneous architecture is not optimal for every application.

With CUDA, we have launched a new computing movement towards heterogeneous computing, where a traditional CPU is combined with a massively parallel processing GPU to achieve a giant step-up in performance.

Visual computing is one of the most well-known parallel computing applications.

There are many others.

Applications such as physics processing, computer vision, video/image processing are just a few examples that are dramatically enhanced by CUDA and heterogeneous computing.

Engineers and researchers in software companies, computer companies and in universities around the world are recognizing the importance of heterogeneous computing.

This is a major movement in computing.

Tesla, our new family of server workstation-class GPUs, with supporting system and software technologies for high-performance computing -- the first Tesla 1U rack contains four GPUs, each with 128 processors, delivering a whopping 1.1 teraflops of sustained supercomputing performance in a real-world application.

Hybrid SLI -- like a hybrid card, Hybrid SLI allows a small GPU and a large GPU to work together.

The small GPU runs like visual computing workloads while conserving power.

When heavy workload is required, both GPUs combine their capabilities to deliver the most delightful experience.

With Hybrid SLI, we can deliver the ultimate visual computing experience while enabling environmentally friendly, power-efficient design.

Lastly, PureVideo HD -- the first video decode and postprocessing technology for Blu-ray and high-definition DVD.

Year over year, annual revenue of each core product line turned in strong results.

Desktop GPU revenue grew 38%.

Notebook GPU revenue grew 114%.

Workstation grew 27%.

MCP revenue grew 7%.

NVIDIA held the number one segment share in desktop GPU, notebook GPU, all of DX9 and DX10 GPUs and workstation solutions.

NVIDIA was named the most respected public company by the members of the Fabless Semiconductor Association for the second consecutive year.

NVIDIA was named Company of the Year by Forbes magazine.

We acquired mental images, the industry's leading photorealistic rendering technology provider.

Mental images' mental ray is the most pervasive ray tracer in the industry.

Lastly, just this week, we announced the completion of the acquisition of AGEIA, the industry leader in gaming physics technology.

We are pleased with our achievement this year.

Particularly, we are very pleased with our strategic position heading into fiscal 2009.

Let me turn the call over to Marv to discuss our financial results.

I will return in a moment to address our growth opportunities for the coming year.

Today, we will be looking at both GAAP and non-GAAP results for the fourth quarter of fiscal year '08 and for the full year of fiscal year 2008.

The difference between GAAP and non-GAAP is primarily stock-based compensation and its tax effect.

In Q4, there was also an in-process R&D write-off from the mental images acquisition.

As Jen-Hsun stated, revenue for the fourth quarter was $1.2 billion, which is an 8% increase quarter to quarter.

The growth quarter to quarter came from desktop GPUs, which grew 19%, and workstation, which grew 9%.

Notebook and MCP were up slightly, at 1% and 2%.

The strength in desktop came from new products such as the GeForce 8800 family.

Comparing Q4 of fiscal year '08 with Q4 of fiscal year '07, revenue was up 37%, led by notebook and desktop.

Notebook was up 124% year to year, and desktop was up 69%.

Workstation grew by 31%.

The consumer business declined 22% for the year.

The strength that we saw in desktop GPUs for Q2 and Q3 has continued through Q4 and showed signs of continuing in Q1, demonstrating the growing importance of the GPU in the PC system.

For the year, we had revenue of $4.1 billion, which is up 34%.

Gross margin for the quarter was 45.7% GAAP and 45.9% non-GAAP.

This is the first time in 13 quarters that we haven't had a non-GAAP gross margin increase quarter to quarter.

During the quarter, we experienced some cost issues with regard to the 8800GT, which we hope to resolve in the next two quarters.

For the year, GAAP gross margin was 45.6%, which is up 320 basis points.

Operating expenses for the quarter were $287 million GAAP and $251 million non-GAAP.

The GAAP number includes a $4 million write-off of in-process R&D associated with the mental images acquisition.

It also includes approximately $32 million of stock-based compensation.

The non-GAAP operating expenses were up $13 million quarter to quarter or 5.5%, and this includes an additional $3 million of operating expenses associated with mental images.

The revenue for mental images in the quarter was less than $1 million.

We continued to aggressively hire during the quarter and ended the year with a headcount of 4985, which is up 376 from the prior quarter and up 902 over the prior year.

Approximately two-thirds of the headcount additions for the quarter and the year were in the R&D sector.

So for the year, revenue was up 34%, non-GAAP operating expenses were up 25% and headcount was up 22%.

The tax rate for the year was 11.5%, which is slightly lower than we anticipated, as the mix of international sales was slightly higher than we had forecast.

This resulted in a GAAP tax rate for the quarter of 8.2%.

For the quarter, the GAAP earnings per share were 42%, and the non-GAAP were $0.49.

For the year, GAAP earnings were $1.31, and the non-GAAP was $1.56.

On the balance sheet, cash and marketable securities was $1.8 billion, down $43 million from Q3 and up $692 million year to year.

In the quarter, we repurchased $178 million of stock, purchased mental images for cash, and also purchased some EDA equipment and services.

The operating cash flow in the quarter was a little over $250 million.

Accounts receivable grew by $114 million, reflecting both the shipment profile in the quarter and the increase in revenue.

The accounts receivable is very current, and there are no issues with regard to payment.

DSO in the quarter was 50 days.

We built $52 million in inventory during the quarter, all of which was in new products.

Inventory continued to be lean at 50 days, up only slightly from Q3.

Depreciation was $37 million in the quarter, and capital expenditures were $70 million.

For the year, capital expenditures was $188 million.

On the outlook, we expect revenue in Q1 to be better than seasonal.

We expect continued strength in desktop GPUs, seasonal weakness in some others, and the result is we believe revenue will be slightly down.

For gross margins, we will certainly try to get back on track of increasing gross margins and would expect margins to be flat to slightly up.

We expect operating expenses to increase quarter to quarter because of the normal Q1 issues of less vacation in Q1 than Q4, higher FICA expense in Q1 than Q4.

In addition, we expect to have approximately $7 million of incremental expenses in Q1 that will come from acquisitions we've recently made.

The result is that we expect operating expenses to increase 8% to 10%.

Since the R&D tax credit has not yet been renewed, we expect there will be an increase in the tax rate until that happens.

If the R&D tax credit is renewed, we would expect a tax rate of 13% to 14% for the year.

Until that happens, our expected tax rate will be approximately 17%.

With that, I'll turn it back over to Jen-Hsun.

Thanks, Mark.

We are entering the era of digital computing.

Applications with bridge graphics are growing with astounding numbers.

Video games is now a $43 billion industry and larger than movie box office receipts and music.

World of Warcraft, commonly known as WoW, is a massively multiplayer 3D online role-playing game with over 10 million active users.

According to comScore, a leader in measuring the visual world, 217 million video game players immerse themselves in online 3D worlds, and that number is expected to grow 17% annually.

Google Earth combines the power of search with rich imagery and 3D terrain and is transforming the way we navigate.

There have been over 350 million unique downloads of Google Earth to date.

Modern operating systems, Windows Vista and Apple OSX Leopard delivers amazing 3D user interfaces.

Combined, the two operating systems have shipped over 100 million copies to date.

Our growth reflects the increasing importance of the GPU in these applications.

We shipped a record 80 million discrete GPUs last year.

Including motherboard GeForces, we shipped a total of 111 million GPUs.

Our annual discrete GPU revenue grew 50% year over year.

Our Q4 discrete GPU revenue grew 80% from a year ago.

Consumer demand for the richest visual computing experience is driving a movement to rebalance the architecture of the PC.

The CPU has become good enough for the vast majority of users.

Meanwhile, a higher-performance GPU will directly result in a dramatically more beautiful and fluid user experience.

PC enthusiasts, gamers and design professionals have known this for some time.

They have long been the strongest proponents and consumers of the GPU.

Let me give you some recent examples from Gateway -- the Gateway P-Series FX, a very thin gaming HD video notebook, priced at $1249 and features a GeForce 8800 GPU and a lower-end 1.6 GHz CPU.

Relative to a notebook with a higher-end CPU and lower-end GPU, the Gateway FX is twice the performance and yet $200 lower cost.

Here is a recent quote from a CNet review of the Gateway FX7020 desktop PC.

"Gateway's FX7020 represents the type of PC we expect to see a lot of in 2008.

It is a $1000 to $1500 gamer that finally has the graphics horsepower to handle the newest 3D PC games.

Its AMD Phenom quad-core CPU isn't the fastest CPU around, although it is quick enough, but the real horsepower lies in its GeForce 8800GT graphics card.

Until now, few PCs in this price range have been able to handle the likes of Crisis, Unreal Tournament 3 and other new PC games with any kind of decent image quality.

This PC and the forthcoming systems like it should finally deliver the promise of next-gen PC gaming to a wider audience."

OEMs and consumers around the world are learning the same thing -- the CPU is good enough.

Investing more on the GPU will deliver a severalfold jump in application performance and experience.

We call this trend the optimized PC design approach.

The balance of CPU and GPU processing should reflect the category of PCs, whether it is gaming PC, lifestyle PC, workstation PC, all-in-one PC or enterprise desktop PC.

We are excited about the coming years.

The era of visual computing and the movement towards optimized PC design approach will make the GPU ever more central to our computing experience.

At the core of that experience is the GPU.

This is the decade of the GPU.

We would be happy to take your questions now.

(OPERATOR INSTRUCTIONS) Tim Luke, Lehman Brothers.

I was wondering, as you guide for the first quarter to be slightly stronger than seasonal, if you could frame for us how you have perceived seasonal in terms of recent percentages, either up, down or -- and maybe, within that, you talked a little about the strength that you've seen in desktop, how we would perceive some of the different segments to look through the first quarter.

Seasonal, to me, is, for our business, is usually down nominally 5%.

I think that the PC in general is down 5% to 10%.

But for us, it's historically been around, nominally, 5%.

So when I say down, I say it will be down less than that.

With regard to the strength in the desktop GPUs, the strength in Q4 and Q3 has been very, very good in the performance segment.

The 8800GT, as Jen-Hsun said, is selling very, very well.

So we would expect continued strength there, so I would say anything in the performance segment of the desktop GPUs is doing very well.

In some of your recent updates, Marv, you had suggested that, while you'd seen strength in the performance end of the desktop area, you'd seen a little bit of volatility in the notebook arena.

I was wondering if you had any recent updates as to how you've seen that.

Then, separately, in recent updates you had said that you had been targeting growth for the calendar year coming of at least in the 20% range, I believe.

I was wondering if you could give some flavor on how you see the different building blocks of that rate of growth going forward.

Notebook in Q4 was up 1% quarter to quarter.

So it certainly didn't decline.

I think that's pretty good performance.

I would expect it to come up seasonality in the first quarter.

With regard to '08 in total, I don't think I have any change to what we have previously talked about.

With respect to the segments within that, do you have any feel for how we should think about modeling the different areas?

Growth in desktop GPUs, growth in MCP, growth in workstation.

Krishna Shankar, JMP Securities.

As you look at your April quarter, you indicated strength in the desktop GPU market.

Do you see broad-based trends there across all segments of the desktop graphics market, or is it more in the high-end enthusiasts and performance-class market?

No, we're seeing growth in the GPU business.

Year over year, our GPU business grew 80%.

And you can't grow that in just one particular segment.

We're seeing growth across all of the segments.

We are seeing increasing adoption of GPUs.

We're seeing simultaneously a bias towards slightly higher-end GPU.

And all of that is consistent with our belief and what I'm sure you believe and most consumers experience today, that more and more of the applications they enjoy are graphically rich.

So I think that people understand increasingly that rebalancing the system so that more GPU horsepower could be dedicated to the experience will result in a much higher performance.

So PC OEMs, consumers, gamers have always known this.

Workstation professionals have always known this.

Design professionals have known this.

And one PC OEM after another are building computers where the CPU is just good enough, and a lot more resources are dedicated towards the GPU, and the products are selling incredibly well because, frankly, they benchmark well, they use well.

It's just a much more delightful experience overall.

Just a follow-up -- in terms of the competitive landscape, do you see -- how would you describe the competitive landscape in MCPs, especially in the notebook market, and also the dual-chip X2 configuration that AMD has come out with recently?

Can you talk about the media products that would be coming out to compete with that?

Let's see, you asked -- I think there were four questions altogether.

So competition in MCP space -- the way we see it is this.

We have the only GeForce motherboard solution for both Intel and AMD platforms.

We're bringing Hybrid SLI to those platforms.

And for anybody who enjoys -- who don't want to sacrifice a rich graphics experience and all the compatibility with all the applications around the world, even though they have purchased an entry-level PC, a GeForce on the motherboard is really the best way to go.

We continue to see success there.

My expectation, our expectation this year is that we're going to grow the MCP business pretty substantially.

A lot of that has to do with the fact that this is the first year where we have GeForce motherboard GPUs for the Intel systems.

So we're expecting that to do quite well.

With respect to the notebook area, my sense is that wherever we don't differentiate substantially, we won't win the business.

If the customers don't care about graphics performance, if it's an enterprise notebook or the competition has a GPU solution just as good as ours, certainly we won't expect to win the business.

But other than that, I think we're going to do quite well.

Overall, I think we're going to have a great year for MCP.

With respect to X2, you would only put two GPUs -- well, let me come at it another way.

If you're going to put two GPUs in a [net-in] card and you deliver the absolute highest performance in the world, the enthusiast that uses that particular PC will certainly tolerate the fact that it's a much larger solution.

But if it's not the highest-performance solution in the world, as in the case of the X2, then it's just really problematic.

There's no market, really, for a product that's larger, louder and not as high performance.

So I think that the GeForce 8800GTX is still absolutely the best DX10 and highest graphics performance GPU in the world and works great in SLI, and we're going to shortly announce or shortly ship three-way SLI.

So multiple GPUs on a net-in card is only tolerated if it's unambiguously the highest performance.

Anything short of that, it's just too clumsy to be successful.

Gurinder Kalra, Bear Stearns.

My first question is on the cost issues you experienced with desktop GPUs in the current quarter.

Can you discuss them in more detail and how you expect them to get resolved?

The 8800GT was ramped probably faster than any high-end GPU in our history.

As I mentioned in my comments earlier, in just four months of production, we shipped over 2 million 8800GTs.

And this is a high-end GPU.

The die size and the number of transistors of the 8800GT is far more than any microprocessor you currently have, Core 2 Duo and otherwise.

So this is a very complex processor.

And yet we ramp it incredibly hard.

We had some manufacturing challenges in the beginning, and we caught it, we fixed it, and now we're going to see far better yields going forward.

But it affected our costs in Q4.

From my standpoint, our focus was delivering the product to the customer.

Demand outstripped supply, so rather than focus on bringing the costs down, we focused on getting the product out.

Secondly, on the AGEIA acquisition -- hope I'm saying that name correctly -- where do you think it enhances revenue opportunities?

Does it mean that you're open to more end markets?

Well, physics processing and bringing dynamics to gaming is an area of a lot of excitement.

Many games already have physics in it, but rudimentary physics.

And people want to put a lot more physics, a lot more dynamics into games.

It just brings the whole game to life.

Now it looks beautiful, and it behaves realistically.

Whether it's water or cloth or motion or particle systems or whatever it happens to be, physics processing in the AGEIA engine happens to be extremely computationally intensive.

It requires just a lot of processors to deliver the type of real-time performance that the games need.

Our strategy is to take the AGEIA physics engine, which has been integrated into tools and games all over the world, and we're going to import the AGEIA physics engine on to CUDA.

You've heard in my comments that CUDA has now shifted to 50 million processors, GeForce 8 series processors, and over the next several years we will ship a few hundred million more.

The ability to port the physics engine on top of CUDA and accelerate the physics is going to add a ton more value to gamers around the world.

Our expectation is that this is going to encourage people to buy even better GPUs.

It might and probably will encourage people to buy a second GPU for their SLI slots, and for the highest-end gamers, it will encourage them to buy three GPUs, potentially two for graphics and one for physics, or one for graphics and two for physics, or any combination of -- any dynamic combination thereof.

So I'm very enthusiastic about the work that we're doing here, and the game developers are really excited about it.

Finally, they are able to get a physics engine accelerated into a very large population of gamers.

So I think this combination between us and AGEIA really kick-started the physics industry.

Brian Piccioni, BMO Capital Markets.

Just following on with the question with respect to the AGEIA acquisition, would you have a timeline where we would expect to start seeing the impact in the actual product offering?

We're working towards the physics engine to CUDA port as we speak.

We're going to throw a lot of resources at it.

I wouldn't be surprised if it helps our GPU sales even in advance of that, and the reason for that is in the end it's just going to be a software download.

Every single GPU that is CUDA-enabled will be able to run the physics engine when it comes.

So we are primarily talking a software rather than physical changes to the internal organizations of the GPU, then?

That's right, because what makes it possible is CUDA.

Every one of our GeForce 8 DX10 GPUs run CUDA.

CUDA is a parallel programming model that is programmable through C.

So we're going to port physics.

The enabler of our engine is physics.

We want to port the physics, the physics engine, on to CUDA and access all the parallel processing capabilities, and in the case of the 8800, 128 processors in our chip.

So instead of running it on only two cores, now we can run it on 128 cores.

So our expectation is that the parallel processing capability will really dramatically enhance the performance.

I guess that you wouldn't agree with your competitors' comments that physics is dead, then?

Why would physics be dead?

David Wu, Global Crown Capital.

Marv, can you split out these various business segments in Q4 of your -- your typical breakout in your 10-Qs?

The other thing I wondered is, can you talk a little bit about the economics of AGEIA?

It sounded like the first quarter will be dilutive, but when could it be accretive?

Lastly, on the situation on 8800, to improve your cost, does it mean that you can still run on 90-nanometer process, just tune it a bit or do something to get your costs down, or do you need to -- are we talking about a potential [dice rank] down to 65 or 55 nanometer to get the required traditional margins that you get on these high-end discrete chips?

Let me start through the business segment breakout that we would show in the Qs.

The GPU business, which includes desktop, notebook and memory, was up 11% quarter to quarter.

The PSB area, which is workstation and some other products, was up 10% quarter to quarter.

MCP, which is our platform products group, was up 2% quarter to quarter.

Then that leaves you with the consumer business, which was down roughly 18% quarter to quarter.

So that's the segment breakout that I think that you were looking for?

Yes, yes.

With regard to the 8800 on the cost thing, it doesn't require us to do anything differently in terms of process technology or redesign of a chip.

It's just a focus on improving the yields and doing the product engineering, which we haven't been able to focus on because of demand outstripping supply.

So it's just hard work, and we'll do it.

It does not require a process change or anything else.

With regard to AGEIA, obviously it will be dilutive because of lack of revenue in the first quarter and probably being able to classify it as accretive is going to be associated with how many more GPUs we sell because of it.

So I'm not worried about it from that standpoint.

Mental images, obviously, in Q4 was dilutive and probably will be dilutive in Q1.

We will see about that.

So AGEIA -- it's not such a big -- not that you need to have a lot of revenue to cover it in terms of incremental hardware you sell.

No.

Doug Freedman, American Technology Partners.

You spend a lot of time talking about the GPU computing opportunity.

Can you spend a little time talking about how a general-purpose computing engine might be required to grow that market and what your view is for your needs going forward of a general-purpose computing engine?

Let me make sure I understand your question.

Your question is, when you say general-purpose computing engine, do you mean the general-purpose application for GPU computing?

I'm referring to whether it's a MIPS processor, a general-purpose computing processor as opposed to a strictly parallel machine.

GPU computing already has a processor.

We believe in heterogeneous computing, and basically the CPU that's already in your system -- it's good enough, it's fabulous.

Billions and billions of dollars of R&D has gone into making that CPU as good as it can be.

And for single-threaded applications, I think that the performance of the CPU is -- it's a miracle of mankind.

It's fabulous.

But they will never scale very well from here on out, and for the type of parallel applications that we believe are really interesting to solve.

So the problem space that we focus on are all the type of applications that require a lot of parallel processing capability.

So we believe that the world needs a heterogeneous computing model where the single-threaded flexible CPU is combined with or collaborates with a massively parallel processor; in this particular case, we call it GPU.

When we introduced CUDA, which allowed the GPU to be programmed with C, and compiled to with a C compiler, we really, really opened up that computing paradigm.

And it's just because developers don't really want to program their problems in Open Geo and Direct3D, and C is a much more natural approach for them.

So heterogeneous computing includes the CPU and the GPU.

In terms of answering maybe a related question about what is the consumer or mainstream application for GPU computing, physics is a really good example of the first one.

Physics processing is already in so many games, as you know, and it's going to be in almost every single game in the future.

All of a sudden, with physics on CUDA, we're able to bring accelerated physics processing now to every PC in the world.

So I think CUDA and GPU computing or heterogeneous computing, however you want to think through it, is -- this year will officially be a consumer-oriented mainstream processing model.

Could you help us understand a little bit of what expectations you guys have for the DX10 chipset marketplace, how soon and how quickly you think the chipset market is going to move to a DX10-based solution, what percentage by the end of the year, if that's possible?

I believe, in order to get Vista certification by the second half of the year, you require DX10.

So it becomes a requirement for Vista starting the second half of the year.

So I think the answer is we ought to get there as fast as possible, and we're trying to get there as fast as possible.

Hans Mosesmann, Raymond James.

A couple of questions.

Jen-Hsun, in terms of the X2 type or class of products that are out there, will NVIDIA introduce a product like that?

And will that establish or become over time the standard at the very, very high end of the market?

We would do an X2-like product only if it delivers performance that is simply not possible anywhere with a single GPU.

But there's no question that a single GPU is a better approach.

It's better use of bandwidth, as the processors collaborate much, much better.

And the overall performance is just unbeatable.

So if we can't deliver the performance with a single GPU, we will do it with double GPUs.

But either way, it has to be the absolutely highest performance.

So you know my preference, and you know we have a lot of evidence and certainly know for sure that a single GPU is the best approach, but if a double GPU can deliver the highest performance on the planet, it will be [excessive].

If it's not, then you're going to have problems.

Just to go further there, what you're saying is that in the near term it's unlikely that you're going to see something like that from NVIDIA, that you'll stick with single solutions in the near term?

We can't talk about future products.

All right.

Well, I understand your preference (multiple speakers)

I'm still selling my current products.

Fair enough.

Hey, Marv, real quick, just share count or an expectation for number of shares in the April quarter?

And that's it for me.

It was around 600 million, I think, as you've seen at the end of the fiscal year.

I don't expect a significant change going out of Q1.

It might be up a little bit, but basically not much change.

That telegraphs the fact that we intend to buy more stock back in Q1.

Fair enough.

Congratulations.

Heidi Poon, Thomas Weisel Partners.

Responding to the earlier question about your growth target for the year, I think you mentioned that it's still around 20%.

So that's seriously above the industry growth rate.

Could you discuss -- of course, the MCP growth is in there.

But are we also talking about maybe some serious share gains or holding off AMD in the notebook area?

First of all, and we're maybe just unsuccessful in communicating this, but I think the entire GPU market is growing.

We're gaining a little bit of share, but it's not significant.

They are single digits, low teens.

The whole GPU market is growing, and the whole GPU market is growing because the computing experience is more visual than ever.

The reason why it's more visual than ever is because more applications than ever before require rich graphics or take advantage of rich graphics.

So we are just selling more GPUs.

And the GPUs we're selling are slightly more powerful all the time.

And so that's really where the growth is coming from.

I think GPUs is our growth engine, and we really believe that this is going to be a big year for GPUs again.

Is that expectation across the different platforms like desktops, laptops, workstations, incremental gains [and tefler] for this year?

Absolutely.

We're expecting growth everywhere.

Secondly, could you comment on, for the growth in Q1 and Q2, maybe the pricing environment and also what you're seeing in terms of any demand differentials in international versus U.S.-based demand?

I don't know the we're seeing dramatically different demand in all the different regions.

If you look at our growth Q4 to Q4, it's 80%.

So I'm not suggesting that Q4 to Q4 will be 80% again.

I hope it will be, but I'm not suggesting that, but the point being that there's some evidence that the consumption and demand for GPUs is increasing, and our expectation is that the growth of GPUs will overcome the seasonalities, whatever seasonalities that there is in Q1 and Q2.

And the pricing environment?

What is the pricing environment?

Pricing is good.

We're seeing more money spent on GPUs, so it has an appreciation in the average selling price for all of the GPU business.

Shawn Webster, JPMorgan Securities.

Good quarter.

Just to follow up on the pricing question, can you tell us what your sequential change in pricing was for your GPU business?

Up nicely.

Up nicely?

Yes.

That's because we shipped a lot of the 8800GTs, as Jen-Hsun said.

So you would expect that.

And so in the various segments, the pricing was good.

We would have had an ASP appreciation because of the higher volume of 8800GTs.

And then, Marv, on the OpEx growth in Q1, can you give us a sense of how you expect your OpEx to grow after Q1?

Internally, Jen-Hsun and I are working hard to keep it flat.

We will see how successful we are.

Like very small incremental changes?

We will see how successful we are.

And then finally, can you talk about the supply environment?

And do you have any of your products where they are seeing leadtimes extended?

Supply is certainly an area of great concern of mine.

We're becoming -- we are the world's largest fabless semiconductor company.

The products that we build are very complex, and the die sizes of our high-end GPUs is significantly bigger than CPUs.

We now sell, as you heard earlier, over 100 million units a year, and it is still growing.

So these processors consume a lot of wafers, and we're already the world's largest.

So it's important that we work very, very closely with our foundry partners.

TSMC has done a wonderful job for us, and UMC has done a wonderful job for us, and my expectation is that we're going to take a fabless model to new heights.

But when you are as large as we are and running as fast as we are, you can never take supply too seriously.

So we obviously take it very seriously.

Are you seeing your leadtimes stretch out or pull in for your graphics or chipsets?

What's the leadtime environment like?

Leadtime hasn't really changed.

Michael McConnell, Pacific Crest Securities.

Looking at the guidance model for better than seasonal for Q1, is this really unique, do you think, to NVIDIA with respect to you had a lot of pent-up demand, obviously, for the 8800GT, supply now is starting to improve, so with that dynamic, that leads to, like you said, some growth in the desktop GPU business?

Is that it, or is there something more here?

Let me take that question.

I think the GPU is just growing.

I think the GPU market is growing.

Frankly, it ought to help anybody who builds a GPU.

So I consider -- the way I see it is visual computing is just becoming ever more important, and so if other suppliers of GPUs are seeing ever-increasing demand, I'm not surprised by that.

It's not to do with -- I don't think we're capturing that much share.

If we look at just the competitive landscape for the fiscal year, can you just talk about what you think from a unit share perspective or a revenue share perspective, probably more importantly, with each of the three main business lines, or even the four, if you talk about maybe desktop GPU, notebook GPU, MCP and then workstations, relative to this last fiscal year?

I don't know that we're that prepared to answer that question.

Not for the full year.

I don't think we break it out quite clearly that way.

Any timeframe, though, that you're comfortable with at this point to discuss, if at all?

Yes, last quarter.

I'm really trying to say this so that we all hear it.

I believe we're growing the market.

The share gains are nice, and we're going to continue to fight for our share and fight for more than that.

But I honestly believe the market is growing.

You can't look at the size of the GPU market by combining the -- by looking at last year's GPU market of all the suppliers and then try to think that this year the opportunity is shifting a few dollars from here and there, because I don't think that makes any sense.

If that was the case and you added S3 and Cirrus Logic's revenues together and Trident's together of when we started, NVIDIA would be a $300 million company.

So obviously, the GPU market in visual computing is becoming more and more important.

What we're seeing here is and what we believe and what we had hoped all along is that the GPU is -- the overall market for GPUs is growing.

So you have to look at -- I frankly think that you have to look at, in order to see the opportunities, you have to look at the entire electronic spend of the PC industry.

There is a vast movement to re-architect the PC, to shift the resources around to deliver the best possible experience and the best possible experience for each market category or each PC category that is being built.

A gamer's PC would have a different allocation of electronic spend.

An all-in-one consumer, multimedia-oriented PC will have a different, another allocation.

So you have to look at all of the electronics dollars in order to decide how fast to estimate the growth opportunity.

But I think share is just simply not that interesting anymore, and certainly, it doesn't explain our growth.

And then just final one for me, on the gross margin, obviously the Company has done a great job through the years of continuing to expand the gross margins.

Taiwan Semiconductor has talked ad nauseam over the second half of last year about trying to renegotiate pricing with some of their larger fabless customers.

I understand the rationale behind what happened with the 8800GT from a cost standpoint.

Is there any type of pricing negotiations that also came into this with respect to the gross margin in Q4, or is it just purely demand exceeding supply and then, as we start to see more balance, the gross margin should start to come back?

The gross margin impact of 8800GT or the increased cost of 8800GT was completely my fault.

It has nothing to do with Taiwan Semiconductor Manufacturing Corporation.

We will have to accept all of that responsibility ourselves, unfortunately.

Now, the good news, of course, is that because it was completely our fault, we can completely fix it.

I think that that one -- let blame where blame needs to go here.

Okay, so this is not anything in terms of pricing negotiations, and this is something that you think we can start to look -- we can continue to expand gross margins as we work through the fiscal year?

You're fairly confident on that?

We would love to blame other people.

Sometimes you've just got to look at yourself.

Tayyib Shah, Longbow Research.

Jen-Hsun, can you please describe the competitive landscape in the notebook space, you versus AMD and then also discrete graphics versus integrated, as you head into the next refresh cycle?

In the notebook segment, AMD has wonderful offerings.

Don't forget that AMD still has a terrific GPU design organization.

So between NVIDIA and HEI, or AMD now, it is unambiguously the two best GPU design teams the world has ever seen.

So I give them a lot of respect.

In the AMD notebook segment, the lower-end segment, it is hard for us to differentiate relative to AMD.

As I said earlier, if we can't differentiate, we don't typically win the business.

So I'm not surprised, and we're not expecting to win any of those.

We are going to fight hard for them, but in the final analysis, I think AMD has good offerings there.

In the upper half of the market, we have quite a bit of differentiation.

Our advantage is that we invest more in GPU development then any company in the world today, and potentially all other competitors combined.

So our R&D investment, our commitment to GPUs, our velocity of innovation -- that is our advantage on the upper half of the market, whether it's notebooks or otherwise.

So our velocity, our commitment and our level of innovation and level of investment is a structural advantage.

So we have to sustain that.

Marv, just curious how you were able to build some inventory during the quarter.

Does that reflect slightly better supply from the foundry, or was this related to demand trends?

No, it's new products in the pipeline.

If you looked at the inventory build, where it actually happened, it's all new products which are just in the pipeline.

Glen Yeung, Citigroup.

Jen-Hsun, can you just be clear as to what you think the PC market is actually doing in this environment?

I understand that GPUs are growing and you may or may not be taking share within that, but just your overall sense of the PC market?

We monitor sell-through, and we are fine.

What we see is fine.

We monitor our own sell-through; I guess I don't monitor other people's sell-through.

But sell-throughs look fine.

We watch it every single week around the world.

It looks good.

I don't really know what else to say.

Fair enough.

Any differentiation with what you see between an OEM basis or a white-box basis?

Not really, not really.

I think that it's possible that the SKUs that we're in are selling well.

It's possible that people are just preferring platforms that we're designed into.

But I don't really know.

I think that overall, overall I believe that the movement -- that two things are happening that are very, very big deals, I believe, that we're experiencing, and we're experiencing them all over the world.

The first thing is the importance of visual computing and therefore the GPU at the core of that.

The second thing is the rearchitecture of the reallocation of resources within the PC, that for different types of PCs you ought to spend and invest your resources in different components.

So the balance between the CPU and the GPU is changing.

I think those two factors combined are really, really important.

I don't believe that to be a one-quarter thing.

I think it's logical that it should be a permanent thing, it should be a forever thing.

I think it happens to cars, it happens to just about any other mature industry, that as the industry matures, you see all of these -- it bifurcates into niches, and the niches become large.

In order to make a niche product specific, you can't use a basic platform.

You have to optimize it for the needs of that niche.

I hear you, and I guess I'm thinking if we can add the product strength of the 8800GT, then one could conclude that even if we were seeing some slowing in PC units at the moment, those trends may outweigh that.

Fair assessment?

Yes, I do.

I think that's fair.

And then just a quick question, Marv, for you.

You had mentioned you have had some cost issues with the 8800 and you are working through it.

In your inventories that you had exiting the fourth quarter, are some of those, the high-cost or lower-margin parts, still in that inventory, and do you expect to work them through in the course of the first quarter?

No.

If we had a cost issue, it means the parts are not there.

It means that they were written off.

So no, they are not in that inventory.

There are no more questions at this time.

I would now like to turn the call back over to Jen-Hsun.

Please go ahead.

Thank you all for joining us today, and we look forward to reporting our progress in Q1.

Ladies and gentlemen, that does conclude the conference call for today.

We do thank you for your participation and ask that you please disconnect your lines.