Shares of Cavium (CAVM) have skyrocketed over the last several months alongside growing enthusiasm surrounding its upcoming Project Thunder line of server/storage/networking chips based on the ARM (ARMH) architecture. Do these prospects make Cavium a buy?

Cavium is expensive, but not unprecedentedly so
Trading at north of 1,000 times generally accepted accounting principles (GAAP) net income, Cavium Networks looks like one expensive chip stock. The problem with looking only at the trailing 12-month price-to-earnings ratio is that high-growth technology companies tend to invest pretty heavily ahead of new opportunities, often masking the earnings power of the businesses that currently generate revenue. 

If you look at what Broadcom paid for network processor vendor NetLogic in 2011 ($3.7 billion for a company that did $405 million in sales at a low-60% gross margin profile), Cavium appears relatively cheap. This is based on analyst consensus for Cavium to reach $439 million in fiscal 2015 revenue.

Given that Broadcom took a $501 million writedown on its acquisition, Broadcom's new fair value estimate should be about $3.2 billion. If Cavium commanded a similar market capitalization, its shares would be worth in the neighborhood of $60.

Is this market viable or simply vaporware?
Looking at ARM Holdings' estimates of what the market for semiconductors that go into the data center will look like by 2018, the picture looks rather optimistic:

Source: ARM Holdings.

Of this $20 billion opportunity, ARM seemingly expects that its partners, in aggregate, could capture about $5 billion-$7 billion worth of chip sales by 2018. The problem for Cavium is the box in the lower left hand corner. How much of the approximately $6 billion in chip value (assuming the midpoint of the penetration estimates) can Cavium really capture given roughly 15 vendors? Note also that a good chunk of the Project Thunder business probably won't be new, but will simply replace the company's current MIPS64-based offerings. 

Furthermore, as is always the case in any research and development-intensive chip market, the number of players here will eventually come down as the market proves unable to support all of them. If Cavium is one of the last players standing, then -- assuming this TAM and a reasonable market share -- the shares look undervalued relative to its long-term prospects. 

Foolish takeaway
It's always exciting to bet on the little guy because the potential for outsized returns is pretty phenomenal. However, the downside is that very often the little guy makes a lot of noise that ultimately amounts to nothing.

Despite the promise shown by Cavium, its first Project Thunder products haven't even made it to first silicon. Even when that finally happens, there's no guarantee that they'll be competitive with the other ARM offerings or, probably more menacingly, the alternative architecture offerings that currently rule the proverbial roost.