cristobal's Blog

Subscribe to this blog

MRVL Blasting Off (The hidden numbers will do the work)

Tags: Semiconductor, MRVL
4 Jan 9:09pm

Marvell Technology Grou

MRVL (NASDAQ)

Forecast has expired Ended: Friday May 11, 2007

Target

$28.38 To $39.62

$14.05 (70.4%)

Time Span:1-3mths

Final

$6.98

-$12.97 (65.0%)

Recommendation

Rating:Buy

Risk:Moderate

Stop Loss Price:Unspecified

Final XP: 0

Target not reached

Analysis

The main thesis of this pick is that Marvell may have booked more sales at Research in Motion and Apple than have been announced because of the hush related to the options backdating probe, which is nearing resolution. How can RIMM and APPL soar and MRVL stay below $22?  My hypothesis is lack of transparency into business outlook. Consumer electronic wins should pass through to component manufacturers. What's so strange is that this has yet to happen with MRVL.  In addition, MRVL didn't move off its lows (like Broadcom has) since it mid-summer decline.  This is because analysts have been bearish on the payroll overhead associated with the acquisition of the xScale processor and 1400+ employees from Intel. In the short term the analysts were right.  However, looking beyond the current FY, if MRVL is able to leverage this new-found application processor expertise into a baseband processor business at RIMM and PALM, then the low end of the market (NOK / MOT ) could be next.  It's a possible direct attack on Teaxs Instruments and Qualcomm's baseband signal processor businesses.

In summary, for the first half of 2007 I like MRVL becuase it hasn't followed it big brother Broadcom half way back up and deserves too given that the company provides better visibility on the xScale costs (ie. is able to show that it has avoided margin erosion) and come out clean on the other side of the options scandal. Longer term, I like MRVL because they may be able to win an vertically integrated set in the cell phone semiconductor business. The have, in a way, bet the company on the xScale, in much the same way that they bet that the disk drive read-channel architecture could be successfully modified for the creation of a Gigabit Ethernet business.  I'm betting they can do it again.