Nvidia's Fourth Quarter: As the Semiconductor Bellwether Turns

A Bellwether of Semiconductors

When Cisco (CSCO) reports earnings, it is one of the first companies to offer insight into how the calendar quarter is progressing for the economy, since its fiscal quarter is one month behind the calendar quarter.

The same might be said of nVidia (NVDA), except that nVidia is more specific to semiconductors. Many semiconductor companies reported a quarter that ended in December, then nVidia’s financial results included January. When nVidia issued a warning last month, it was a heads-up that January was shaping up to be a bad month too for the industry.

Most people seem to have overlooked the single most important fact for the semiconductor industry to come out of nVidia’s last fiscal quarter. Inventory in the supply chain has at last contracted to the point where the customers of semiconductor designers should start placing more orders again soon.

The problem so far, for semiconductor companies and many others, has been a reset in end-user demand and consequent glut of inventory in the supply chain. The inventory in the channel needs to burn off, so the semis can start taking orders again that are more reflective of end-user demand. The big news out of nVidia is that “inventory levels are depleting very quickly” (Earnings Call).

Taking excess inventory out of the supply chain does not mean we are out of the woods, but it is an important step. End-user demand is still weak, but it’s not as weak as the financial results nVidia and others reported last quarter.

Graphics Processors [GPUs]

Eyebrows must have been raised when nVidia warned in January that revenue would be down 40-50% sequentially. The reason why this warning was surprising was there were cogent reasons to think nVidia was doing well in the competitive marketplace.

Market share seems to have accelerated with, of all things, a driver update. Big Bang II gave nVidia a significant edge in performance on most new games. In particular, nVidia’s second best GPU, which competed head-to-head with AMD’s best single GPU, the Radeon HD 4870, pulled ahead often by wide margins.

The release of the drivers appears to have been well timed. It seems nVidia was holding back on releasing the drivers until shortly before the Christmas buying spree. As one website put it after benching the two cards, “Be sure when you set [sic] on Santa’s lap you tell him no GTX-260 Core 216 or don’t slide down my chimney!” (Bjorn3d).

Possibly most important, nVidia launched the GeForce GTX 295 in early January. With the GTX 295 nVidia took back the single graphics card crown. The best graphics processor and the best graphics card are not necessarily the same. Both AMD and nVidia put two graphics chips on a single card to try and make the best overall performing card. It’s a halo product, probably carrying significance far beyond itself. Whoever has the best single card, no matter how it is made, is often perceived to have the best graphics chips.

Also important, nVidia improved the best single graphics processor on the market with the GeForce GTX 285.

The result was nVidia took market share in calendar Q4 and almost certainly in nVidia’s fiscal Q4 too. Revenue was still sharply down, but market share was up. Consequently the CFO said: “we believe we regained some market share particularly in the high end”.

So we had a good idea that nVidia was doing well competitively, and yet, what to make of the news in mid-January that revenue would be down 40-50% sequentially?

Investors had lost sight of what they knew: inventory in the supply chain was contracting, and nVidia has exposure to a lot more than just desktop graphics.

In the event desktop graphics did better than everything else, down 34%.

Notebook graphics, a high-growth segment, were down almost twice as much at 63%. Is this cause for alarm? Not necessarily. We know that the supply chain for notebooks is very long, so there is more inventory to take out, and we know that inventories now are lean.

The only uncertainty in my mind is how much of the drop off in workstation graphics was inventory correction, and how much of it was share loss. AMD has a good chip, nVidia had almost all market share, so one would expect AMD to take some share.

Revenue came in right where nVidia warned, 46% down. To be sure, there were other things in the quarter that could have been better such as nVidia’s own inventory. Nevertheless, if analysts are upset, it seems primarily because they feel nVidia is spending too much money.

Investing Out of the Recession

The same day nVidia reported earnings, Intel (INTC) made an important announcement that morning. Intel would spend $7 billion investing in its factories in the U.S.

The investment will help Intel maintain its lead in process manufacturing, possibly even extend that lead. However $7 billion in spending should also provide a boost to the economy. The president of the U.S. is alleged to have said “it was the first real piece of good economic news he had received since taking the oath of office” [CNBC].

Intel’s CEO exhorted others to do the same, invest in the downturn to help lead the country and the world out of the mire it is in: “we want to lead by example here and ask others to follow” (Forbes).

Other companies were unlikely to follow suit, suggested one headline (Dow Jones).

The announcement of $7 billion did little to help Intel’s stock, which dipped as much as 7% that day. I would ordinarily chalk up such a dip to coincidence if it were not for the hard time analysts gave nVidia later that day for doing the same thing Intel was doing. nVidia’s stock slumped as much as 15% the following day.

I couldn’t help but be struck by the irony. Banks and Wall Street greed are what got the world in its current mess, and here were representatives of these institutions frowning upon companies for doing something that could help recovery, not to mention which was good for the companies doing the investing.

A Little R&D

nVidia’s most damning technology critic has suggested that there has been a departure of engineering talent from nVidia: “the good people are now at Intel” (The Inquirer). Against this, Jon Peddie suggests AMD and nVidia “have probably 4x the number of world class graphics engineers that Intel has” (Hexus). I don’t know which of these views is closer to the truth. But one thing is certain. nVidia is using the recession to beef up its talent.

Warren Buffett is buying stocks in these dark times. To the CEO of nVidia, however: “It’s a great time to recruit” (VentureBeat).

Headcount has increased by 127 sequentially and by 435 year over year. Humorously these increases come with a hiring freeze in place and “with only the most critical positions being filled” (Earnings Call).

The recruiting appears to be a part of a much larger investment in R&D. The company is spending more this year than “last year — by a lot” (VentureBeat).

Analysts’ Reactions

Admittedly this recession is not making the jobs of financial analysts easier. For one, seasonality has been thrown out of the equation, and analysts still have to predict results.

Q1 is usually not as good as Q4, because Christmas and other holidays are in Q4. So one would normally think revenue for Q1 would be down from Q4. However, the contraction in the supply chain has thrown all such adjustments into question. The CEO interrupted one analyst: “there’s nothing that we’re experiencing right now that is seasonal” (Earnings Call).

Instead of accounting for typical seasonality and guiding down from the fourth quarter, management observed that the company’s selling into the channel was less than the selling from the channel. At some point soon they should start to see more orders. nVidia consequently guided flat to up on sales, though they don’t know “when and by how much revenue will increase” (Earnings Call).

Revenue projections being what they are and expenditure estimates being what they are, it does not appear nVidia will break even in Q1. nVidia is making cuts in spending; however these cuts are not based on trying to meet a breakeven point. Management was quick to point out to one that nVidia is cutting Op Ex not based on a breakeven but on “what we can do without jeopardizing future revenue” (Earnings Call).

Just so you’ll know, nVidia ended the quarter with $1.26 billion, which was down about $49 million since last quarter, so it’s not like they are burning through cash at an alarming rate. They will burn through some cash, though.

It has been said so many times that it sounds stale, you invest your way out of a recession. No one however has said it better than Marvin Burkett, the CFO of nVidia.

One analyst went so far as to say that things were so bad that future growth wouldn’t be hurt, so why not cut spending more? Sigh. The CFO replied that he did not know how long the recession would last. However, there was one thing he did know. “I do know the only way out of recessions and significant recessions like this is not by cutting your future. The way out of the recession is by investing in future revenues” (Earnings Call).

Larrabee

One of the reasons nVidia needs to heavily invest in R&D is they have a tiger by the tail. Intel is set to enter the fray of discrete graphics processors.

It doesn’t help relations between Intel and nVidia when Jen-Hsun Huang, the CEO of nVidia, says things like “the GPU is the new soul of the PC” (Earnings Call). In the latest spat between the two companies, the CEO penned, “the CPU has run its course” (nVidia).

Intel will soon leverage its advantage in process technology to deadly effect. nVidia will oppose Intel’s edge in process technology with product technology. However, Intel is smart, and even Intel knows not to directly take on nVidia head-on in graphics. Larrabee is a very different animal. Intel does not even like to refer to Larrabee as a graphics processor; rather it’s a visual processor.

Larrabee will have achieved its end if it can establish a toe hold in any part of discrete graphics. Intel is steady and methodical in what it does and is not in my opinion aiming for a knock-out blow. At least not at first. If Intel can achieve a measure of success with such a radically new part, perhaps Intel and nVidia can both grow the market in different directions. I hope so. The industry needs competition.

For the moment, however, nVidia appears to be taking seriously Intel’s entry into their business and is preparing accordingly.

I’m sure that Abu Dhabi would love for Intel to stop investing in process technology throughout this recession. Because Abu Dhabi will not stop investing in AMD’s fabs.

Similarly I am sure Intel would love for nVidia to stop investing in visual computing. For Intel is not going to stop investing in it.

Maybe nVidia is not spending enough?

Disclosure: Long NVDA and INTC

Comments

[donzelion:]

Cisco will tell you more about big-ticket macro developments; nVidia will tell you more about consumer tastes and spending capabilities.

I share your ill-ease with investor response to Intel's $7 billion plan for investment in U.S. manufacturing. Of course, since many of the banks that were screaming "buy buy buy" during the peak are now screaming "sell sell sell" - one ought to take them with a grain of salt (as for me, I prefer to just change the channel and go with what common sense suggests - Intel is a powerful company without peer...)

As for nVidia v. Intel - I see Intel's moves towards GPUs as a backup strategy in case it grows too cost prohibitive to continue with the die shrink portion of their tick tock strategy. It's not that Intel is somehow more methodical than other semis - it's that they are strong believers in backup plans after Pentium literally burned them. Besides, Intel wants nVidia to stick around (albeit, perhaps in an ever shrinking niche).

Feb 24 09:33 AM

[principle:]

Intel doesnt have to make a great graphics card for it to be a success, thats the beauty of it. Intel can just make it and sell it cheaper to system builders for all of the baseline systems that want something more than an IGP, or additional display ports. Its beautiful for Intel to increase margins for each PC sold, while squeezing Nvidia out of the low end to mid market completely, since AMD isnt going to use Nvidia cards. Nvidia is also being completely squeezed out of the IGP market with Nvidia not allowing Nvidia to build chopsets for Nehalem, so no more Nvidia IGPs in Desktops or laptops, unless Nvidia starts making their own systems with better processors than their using in their netbooks now. Nvidia can have the performance crown for the ridiculous amount they charge for it, but they are going to need to come up with a really good strategy very soon or they are going to lose half of their business.

Feb 24 10:50 PM

[Misha:]

I found that the following recent IEEE Spectrum article provides a concise and accessible description of why Intel’s (INTC) Larrabee may be such an important product with consequences with for NVDA:

www.spectrum.ieee.org/...

Feb 28 06:34 PM

[Phil Coffman:]

A very good article. Thanks for pointing it out.


On Feb 28 06:34 PM Misha wrote:

> I found that the following recent IEEE Spectrum article provides
> a concise and accessible description of why Intel’s (seekingalpha.com/symbo...)
> Larrabee may be such an important product with consequences with
> for NVDA:
>
> www.spectrum.ieee.org/...

Mar 01 11:22 AM

[Misha:]

Phil,

And while they finish Larrabee, they do whatever they can to slow NVDA:

www.marketwatch.com/ne...={D6B97049-D319-45B2-9...

I went long INTC some time ago (too early, as it turns out).



On Mar 01 11:22 AM Phil Coffman wrote:

> A very good article. Thanks for pointing it out.

Mar 01 12:11 PM