Your Best Chip Returns Are Absolutely Fabless

by: Dana Blankenhorn

Everybody knows Moore's Law. Chip densities double every 12-18 months. The amount of bang for your buck you get from chips gets better-and-better, faster-and-faster.

But few understand Moore's Second Law. It's a corollary to the first. It states that as chip densities double, so does the capital demands for producing them.

Because of Moore's Second Law, chip foundries have become increasingly capital intensive with time. The result has been the rise of fabless semiconductor companies, firms that mainly design and market chips, leaving the production to someone else.

Today the hot place to be is fabless. Qualcomm (NASDAQ:QCOM), Broadcom (BRCM) and Nvidia (NASDAQ:NVDA) all now sport PE multiples ranging from 15-20 while Intel (NASDAQ:INTC) and Texas Instruments (NYSE:TXN), the last of the integrated chip companies (AMD having split its foundry into the company GlobalFoundries in 2009), now carry PEs of around 10.

The problem is inherent in the structure. The cost of production is driving the train at both firms. They go about the business backwards. A fabless company like Broadcom thinks of products, not designs. It sells entire reference designs to OEMs rather than just chips. It's closer to the market, and thus more profitable.

Take a look at Nvidia, which is rising today but actually trails its fellow fabless firms in valuation. It has gross margins of about 50% of revenue, takes 15% of its revenue into operating income, and sports a balance sheet rich in cash. Intel starts with similar margins, but operating margins have been falling for years, and net income has stalled. Less than half its assets are in cash.

Slowly, but now obviously, Moore's Second Law is doing its work. Separating design and marketing from production, even on a merely internal, functional level, can start to cure this. Let the factory guys seek production work where they can get it.

Focus design and marketing on OEMs and products, rather than on chips and production schedules. Over time, make the split increasingly obvious and then, when the market is right, divide the whole in twain.

Until something like that happens, at least internally, know that your best chip investments are absolutely fabless.

Disclosure: I am long INTC.

Additional disclosure: Yeah, I know, my portfolio ignores the point of the story. Maybe I'll be making some changes soon. But I haven't made any firm decisions.