Marvell Technology Group (NASDAQ:MRVL) has been quite a wild stock over the past several years. The company's shares gained real momentum a couple of years back, as it was viewed as a hot smartphone component play. However, its core storage silicon business suffered significant setbacks as PC demand waned, and it faced significant legal troubles. The stock crumbled.

Like a phoenix rising from the ashes
But after bottoming in late 2012, the company's shares have been on fire. The PC market is recovering (and Marvell's share position in hard-disk drives as well as solid state drives continues to improve), and its mobile efforts have begun to drive significant revenue growth after a multiyear period of investment. Though the networking market has also been on fire, Marvell appears to be losing share to Broadcom (UNKNOWN:BRCM.DL), among others.

According to sell-side consensus, the company is on track to enjoy 14.20% revenue growth for the current year and low double-digit earnings-per-share growth. That growth is set to slow per consensus to just 3.60% in the coming year, although that sure beats the revenue declines that the company saw in 2012.

The shares look cheap -- what's going on?
Going further with analyst consensus, you'll see that the shares trade at about 13.6 times this year's earnings-per-share consensus. The stock looks even cheaper if you take into account that the company has about $4.22 a share in net cash per share. What gives?

Well, there are two things that probably depress the shares. First and foremost, Marvell lost a patent suit, in which the company was found guilty of infringing upon two patents filed by Carnegie Mellon University, and has been ordered to pay damages on the order of $1.54 billion -- most of its cash pile. Though Marvell is still vigorously appealing this decision, the uncertainty here does depress the valuation.

Atop these concerns, Marvell seems to be chasing fairly low-margin, relatively low-barrier-to-entry. low-cost mobile chip platform solutions at the expense of its higher-margin networking business.

The risk of betting big on mobile (at the expense of other, higher-margin businesses)
Interestingly enough, Marvell and Broadcom seem to be diverging. Broadcom (not so gracefully) bowed out of the cellular applications processor market but has been winning in infrastructure and networking. Marvell, on the other hand, has been seeing success in the cellular apps processor market, but its network silicon division continues to dramatically underperform its peers, with sales down 5% in fiscal Q4 2014 and 3% in fiscal Q1 2015, respectfully. 

This is a bit of a risky play. The low-end mobile applications processor market is really growing fast, and given the right scale and cost structure, this can be a very lucrative business -- just ask MediaTek. However, for a company like Marvell, which still isn't at scale, and given that Marvell's recent success could come under threat from MediaTek (which, broadly speaking, is at scale) when it rolls out its own 4G LTE solutions, Marvell is in a pretty risky spot.

Further, with Samsung (NASDAQOTH:SSNLF) -- one of Marvell's key LTE clients -- having announced a competent integrated apps processor and modem solution for the low end, the pressure there will only continue to mount.

Foolish bottom line
It will be interesting to see whether Marvell will be one of the few remaining survivors in the mobile applications processor space as the space eventually consolidates. If it can cross that chasm and be one of the few profitable players, then this "bet" will have been worth it. However, if it ends up pulling a Broadcom, then it will not only be out of this growth opportunity, but it will also potentially have left meaningful and fairly high-margin networking business on the table.