***way OT***
>have been watching that one go up for years, never bought it though, cause everyone thinks that GaAs is just a temporary stopgap technology (ie, before we get Si-based compounds), and thus vitesse a short-lived company.
thanks, kas. you pose a valid concern. i think that and the fact that vitesse got its jumpstart as supercomputer supplier to the likes of CRAY, gives people the impression that they're a white dwarf.
if you're interested, some thoughts i've previously posted (in various forms) to the VTSS board on this topic. if not, my apologies.
-----
for the record, IBM's Microelectronics Division pioneered SiGe (silicon doped germanium) technology. as you may know, george gilder has previously forecasted that GaAs will eventually succumb to SiGe. most of the evidence is to the contrary: GaAs is still too difficult for many companies to master due to the inherent instability of arsenic, which causes threshold voltages to vary across the wafer, thus lowering yields. and unlike silicon, a high percentage of the cost for GaAs is the wafer itself, so the barrier to entry isn't insignificant. in addition, as you can imagine, VTSS has its own proprietary manufacturing technologies, yet utilizes industry standard fab equipment (e.g., standard AMAT metal- and chemical-vapor deposition equipment).
SiGe promised equal to better performance at cheaper costs, but production hasn't bared that. proponents of SiGe say they can get all the individual transistor speed of GaAs, but with far less noise, much higher uniformity of performance across the wafer, and greater thermal conductivity. we'll see ...
if you're looking at competition, PMC-sierra excluded, the company that most often comes up as a VTSS competitor is applied micro circuits (AMCC), which signed a deal in july to use IBM's SiGe technology for both SONET and ATM applications. VTSS and AMCC even share a few customers: alcatel and cisco come to mind. however, if you look back, history has already shown that many bipolar silicon players have opted out of the market (e.g., AMD, fujitsu, NEC, and texas instruments). so while GaAs isn't exactly easy to produce, it's now arguably become a simpler manufacturing process. for VTSS, anyway. witness their new facility in colorado springs -- world's first 6-inch GaAs fab.
i also think it's significant to note that the big telecoms who buy these integrated circuits don't switch suppliers lightly. VTSS has locked up a number of customers on GaAs and the switches lucent, ericsson, et al. make from them don't exactly come cheap.
another notion to contemplate: VTSS exhibits classic symptoms of guilt by association. for example, it's often lumped in with the legion of other GaAs chip makers like anadigics, alpha, celeritek, remec, and triquint, so when these companies dip, VTSS has been known to slide too. however, these other outfits sell linear low-noise, high gain GaAs chips for the wireless cellular handset and infrastructure market; VTSS, on the other hand, sells high-level integration digital GaAs chips for the data- and telecom, and networking markets. in short, they all use GaAs, but VTSS participates in a different market and sells very different chips.
from a business perspective, two customers, lucent and schlumberger represented 23% and 15% of sales, respectively, in FY98, up from 20% and 12%, respectively, in FY97. i guess you can read that both ways. also, VTSS's tax rate is increasing. in FY97, VTSS paid a 10% tax rate due to large operating losses carried over from earlier years of R&D. FY98 was a mix -- overall paying taxes at 20%. according to tomasetta, FY99 should be in the low 30%. |
