Friday, May 25, 2007

Marvell: Hero or Zero?

Marvell Technology Group (MRVL) was a hero to its investors back between mid-2002 and late 2005, when it rose from under $5 per share to over $30. Those who bought in too late (including me) have been disappointed by its slump to near $15 per share. Is Marvell now a bargain, a hero merely stunned in battle, or is it wounded and heading for the trash bin of semiconductor history?

Marvell makes semiconductor chips and is noted for advanced chips integrating digital and analog functions. The chips go in switches, transceivers, power management, printer, and communications products. But above all they go into hard drive based data storage devices.

If you look at revenue on an annual basis you will see rapid growth from 2001 ($288 million) forward to 2005's $1670 million. In the latest quarter report, in the typically seasonally weak Q1 2008, revenues were $635 million, which puts Marvell at an annual run rate of $2.5 billion. What is not to like about that?

First of all, we can only guess what net profits, if any, Marvell has been pulling in. It is one of those stock-option accounting, can't tell you anything until it is done firms. Management did say that, yes, there was incorrect dating of options; the investigation is essentially complete; they hope to file restatements and late statements with the SEC as soon as possible.

The other big shoe that fell was in the data storage market. Western Digital and Toshiba each amount for over 10% of Marvell's revenue. Marvell's superior products now command about 65% of market share. Marvell's rapid growth, and rapid stock price appreciation, came from going from 0% to 65%. Going over 65% is not easy for two reasons: competing chip makers are doing everything they can to hold on, and the hard drive manufacturers want to have alternative sources available because competition keeps prices reasonable.

After that the biggest concern other than a recession or potential weakness in the overall semiconductor market is in the XScale/cellphone sector. In 2006 Marvell bought Intel's division that produced XScale microprocessors that are meant to be used in advanced (3G) cell phones and PDAs. As everyone knows the PDA market is basically over. Intel was losing money in the division and needed to focus on its battle with AMD. Marvell payed too much for this asset, I think, but its executives believe that by combining XScale with Marvell's analog expertise (never an Intel strong point), lowering manufacturing costs, and gaining market share, this division could be very profitable. It is possible, of course, but the competition for chips going into 3G cell phones is intense, with some formidable players like Motorola in the mix. Just getting this segment to break even will impress most analysts.

Much of Marvell's revenue growth these last 2 years has been by acquisitions, including the printer chip division, which Marvell reports is doing well. Storage growth is limited by growth of the storage market itself.

There are a couple of bright spots on the horizon, notably advanced video processing chips. But so far these are contributing insignificant revenues.

The real fear: when the veil comes off, when the restatement is made, earnings will suck due to costs associated with the XScale acquisition. These should be one time costs, but they will give a bad impression.

My take? Worth the risk at this price. Once the accounting disruption is in the rear-view mirror and the Intel acquisition is digested we'll see what Marvell is doing and where it is heading. Marvell has always been well-managed, so I think the earnings picture will be pretty healthy.

Would a U.S. recession matter to Marvell? Some, but it sells its chips into a global economy. With Germany and China booming, with India and Japan coming along, I don't see a global recession near-term.

Reminder: I own Marvell stock.

More information:

My Marvell page, with links to my summaries of analyst conferences
Marvell web site

No comments:

Post a Comment