Marvell's Huge Research and Development Budget

Marvell (MRVL) reported fiscal Q2 2008 earnings yesterday (August 23, 2007) with a GAAP loss of X or x per share. The stock lost 10% of its value in trading today. I believe some traders were hoping for a short-term pop because of rumors that a hedge fund had taken a position and that Marvell produced one part for Apple's iPhone cell phone. They overlooked the elephant in the room: Marvell's R&D (research and development) spending for the quarter. In fact the short-term sell-side analysts probably see the R&D budget as a negative. Here I'll focus on the future and the R&D expenses. For more on the past, see my notes on the Marvell analyst conference.

Marvell spent an astonishing $236 million on R&D in the quarter. Its revenues for the quarter were $657 million. So R&D was 36% of its budget. Compare that with, say Intel, which spent 30% of its revenues on R&D last quarter; or a more direct competitor, Linear Technology, which spent $47 million on R&D, or 17.5% of revenue.

It looks to me like Marvell is girding for war. They are not going to try to become the world's sole supplier of semiconductor chips, but they are going after a lot of cutting-edge, high volume, high margin business.

Some results are already rolling in. In May management guided Q2 revenues to $645 million; instead they came in at $657. Part of the differential probably came from chips that went in iPhones, but you have to remember that Marvell is already a big company. In fact it has pretty well made its intention clear: it is going after the 3G high-end phone market, not just Apple's tiny fraction of it. Last year it bought Intel's communication processor business (which this year is resulting in $36 million per quarter non-cash write offs of amortization costs, which really skews GAAP EPS downward). It was already developing its own communications processors. It also has some of the best analog radio technology in the industry. What it is doing is combining all of these functions - digital signal processing, general digital processing, and cell radio signal transmission with Wi-Fi and bluetooth. That is a killer combo. It does not mean there is no competition, or that any cell phone maker is required to adopt it. But those who do not may find themselves, in a year or two, at a severe competitive disadvantage.

Most companies that have done so well in making chips for hard-drive storage would see taking on the entire cell phone semiconductor industry as challenge enough. But not Marvell. It is also going after some other big hunks of business.

One is video processors for large screen TVs. Again, Marvell came to this game late. Again, the already have a foot in the door with some advanced silicon that makes for better pictures. Again, they are planning to solve a bunch of unsolved problems and deliver a solution that will be irresistible to the TV makers.

In parallel with that they are working on the DVD/ HD-DVD end of things as well.

Oh, and not satisfied with mere dominance in the hard drive industry, they have sunk a bunch of effort into R&D to make even better drive chips; they are confident they can gain revenues in this area.

Then there are their advance in LAN technology. And power management. And printer technology. I'm probably missing something, but that seems to cover the basics.

After a year in which Marvell and many other semiconductor makers got hit by inventory adjustments, slowed demand growth, and pricing pressure, Marvell's intense pursuit of new markets may be bearing fruit. Management believes Q3 revenues may come in around $710 million, up $53 million sequentially from Q2. I'll be very impressed if that happens.

More important, if they are right about design wins they are getting this year leading to significant new revenues in calendar 2008, I can't wait to see what Marvell looks like a year from now.

As to pricing the stock, you'll just have to choose your theory. I don't believe in buying stocks with high PE ratios, and right now if you use the GAAP numbers the PE ratio of Marvell is infinite. If you use non-GAAP numbers, it is pretty high. On the other hand I understand the value of R&D. Another company, trying to please short term investors, could have simply cut back its R&D budget in Q2 and shown some very impressive results. But as a long-term investor I'd rather have the company with the bad short term results, especially when its R&D arm has a record of paying off, as Marvell's has.

Right now I already own as much Marvell stock as my portfolio model allows, so I won't be buying more anytime soon. Even holding it at this level involves risk: just because Marvell is spending money on R&D does not mean it can sell any resulting technology. But that is a risk I can live with.