mmm@uunet.UU.NET (07/24/90)
From: <ames!ames!claris!portal!cup.portal.com!mmm@uunet.UU.NET> A few months ago, the head of DARPA (Craig Jones?) was removed from his post for a revolutionary act: using DARPA funds as venture capital for promoting strategic technology. An investment was made in Gazelle, a vendor of gallium arsenide IC's. Gazelle has two product lines, superfast versions of popular programmable logic devices, and a gigabaud serial I/O receiver/transmitter chip set. The latter has 40 bits of parallel data strobed into the transmitter, which sends the data at 1.25 Gbaud over a serial link to a receiver which strobes out the same 40 bits. Apparently, the Administration considers this to be meddling in the free market. Several days ago, the San Jose Mercury-News, in its business section reported the following: "Gazelle Microcircuits Inc. of Santa Clara named Anil Bedi director of strategic marketing and Kadiresan Annamalai senior applications engineer. The appointments were made possible by recent funding the company received from the Defense Advanced Research Projects Agency. Bedi was director of marketing at Oki Semiconductor. Annamalai was previously on the technical staff of Advanced Micro Devices." For what it's worth, I think the PLD product line is a loser. AMD and TI have PLD's of about the same speed, without using GaAs. Plus the fact that Gazelle's PLD's must be programmed at the factory (they are burned with a laser). On the other hand, the gigabaud serial I/O chips look pretty interesting. It's also worth noting that Gazelle doesn't have their own fab. They use TriQuint as a foundry. IMHO, a better investment in infrastructure would be TriQuint. One neat way to do that (which is already being done) is to fund research projects which require GaAs foundry services. It seems to me that better uses can be made of DARPA funds, rather than hiring marketing personnel for Gazelle.