Who says the components world is all doom and gloom? Officials at transceiver firm Fiberxon Inc. say business is strong enough for their company to consider going public early next year.

"We're ready to go to an IPO with 2006 revenue projections," CEO Li Hsu says.

That's in contrast to the glum financial situation of Avanex Corp. (Nasdaq: AVNX), Bookham Inc. (Nasdaq: BKHM; London: BHM), and JDS Uniphase Corp. (Nasdaq: JDSU; Toronto: JDU) -- a picture that isn't getting better at the moment (see Avanex Cuts Revenue Forecast). But to Fiberxon and other small players at the OFC/NFOEC 2005 tradeshow last week, the market is looking bright, partly because the newer companies aren't burdened with the expensive infrastructure of the larger players.

Fiberxon sells transceivers to the datacom and telecom markets, in the latter case reaching speeds up to OC48 (2.5 Gbit/s) with an OC192 (10 Gbit/s) generation en route, Hsu says. The product-line makup is a lot like that of Finisar Corp. (Nasdaq: FNSR), he says.

Fiberxon has been profitable for nine quarters running, with revenues that tripled last year, Hsu says. The key has been that Fiberxon operates in China, keeping costs down, and has avoided building up infrastructure. "To make money you have to have a low operating cost," he says. "There's not enough barrier [to entry] to justify high operating costs in the U.S. and Europe."

Table 1: Fiberxon Revenues

Fiberxon doesn't build its own components. Rather, the company buys the pieces and makes its money by putting engineering expertise into the finished modules. So, while the company operates in low-cost China, Fiberxon isn't hiring cheap assembly workers; half the company's 350 people are engineers.

Aside from cost, the model works because Fiberxon can crank out new designs quickly, Hsu says. The company was asked in March 2004 to develop a GE-PON (Gigabit Ethernet Passive Optical Network) transceiver in three months. Fiberxon took about 30 engineers, moved them to a skunkworks location in Shenzhen, away from the distractions of home, and had them hammer away at the part, working 15-hour days without weekends.

Having worked in the U.S. for 20 years, Hsu consciously modeled Fiberxon after a U.S. startup, down to incorporating the company in the United States -- a factor that might make the company a more comfortable play for U.S. bankers. "The whole company is modeled after a Silicon Valley company," down to the stock options in each employee's pocket, Hsu says.

Fiberxon isn't the only tranceiver company claiming good revenues. Privately held Opnext Inc. is rumored to be pulling down $100 million per year, although officials there won't confirm or comment, nor will they hint whether Opnext is profitable.

Unlike Fiberxon, Opnext owns a fab, but the company has added a line of DVD lasers to keep the production lines busy. That's given Opnext a brighter outlook than publicly held firms that are burdened with fab expenses -- although because Opnext is private, it's difficult to tell what shape their spending is in (see Components Competition Is Killing).

"It sounds so simplistic, but that's the main thing -- you have to keep up volumes," says Michael Chan, executive vice president of Opnext.

Fiberxon, too, may have to tangle with fab issues someday. "We have very little fixed cost, and that model worked so far," says Hsu. "But as we grow bigger, I think the business justifies picking up some of the pieces of vertical integration."

— Craig Matsumoto, Senior Editor, Light Reading