To configure or reconfigure chips, that is the question. Whether tis’ nobler to invest in fabless application-specific integrated circuits (ASIC) or to reconfigurable communication processors (RCP) has yet to be determined.
“I think they are nuts! ASIC chips are dead,” said Chuck Fox, president and chief executive with RCP developer Chameleon Systems. “They may not die this year, but they will die in this decade. So, anyone that is investing heavily in new ASIC companies has got a screw loose.”
Such an assertion, however, came close to blasphemy for Jack Harding, president and chief executive with ASIC provider eSilicon Corp. “If you are trying to accomplish the functionality of increased complexity you must have custom designed chips, or ASICs,” Harding argued. “What [Chameleon] is trying to accomplish represents approximately 5% of the market in total, and to describe ASIC’s as a dying market or obsolete technology is ill-informed at best and ridiculous at worst.”
Early in February both companies escalated the competition by closing on a pair of sizable private equity investments. eSilicon polished off its $20 million second round of financing led by Tarrant Venture Partners, while Chameleon closed on its $47 million series D round, led s 3i Group. Both firms are floating the idea of initial public offerings for late next year.
While the rhetoric between Fox and Harding is largely a product of professional pride, there is a legitimate controversy over whether custom-made or reconfigurable chips will ultimately prevail in the New Economy.
On one hand, Chameleon believes that it is addressing a paradigm shift in the way that chips are going to be designed, stating that the equipment companies cannot rely on actual customized chips because they are too expensive and inflexible.
“What we do instead is produce a chip that they program and change,” noted Fox, “But the chip is the same for everyone, which alleviates time and manufacturing issues.”
On the other hand, eSilicon believes that the problem with reconfigurable chips is that the applications are so specific that an enormous software burden will surface when shifting the customization of the hardware into the software.
“The real decision becomes are you better off injecting the specificity by way of the hardware and using generalized software, or are you better off to use generalized hardware and having to carry the burden of writing software,” Harding said.
And the Investment Goes to…
When Crosspoint Ventures incubated eSilicon in 1999, it formed the opinion that substantial changes in the electronic manufacturing industry were taking place.
Seth Neiman, managing partner at Crosspoint, concedes that ASIC may not be the latest technology, but that the semiconductor industry doesn’t consist of the latest technology. Moreover, he believes that there are many different segments within this space.
“The ASIC business is a $12 billion to $13 billion business and to say that it’s fading is a brain-dead point of view,” he said.
3i, on the other hand, discovered that problems concerning the lack of enabling silicon for the next generation of services to be rolled out needed to be addressed.
“The makers of the equipment have two choices today: they can either go and do custom chips (ASICS) that take about nine months to design which is very old fashioned,” said David Silverman of 3i. “Or they can go with programmable chips and Chameleon is coming out with chips that tweak applications and distinguish products.”
What the Future Holds
Despite their differences, both companies feel their greatest upcoming obstacle is managing future growth.
Chameleon plans to double the size of its company, and eSilicon may well expand given that it believes revenue generation is right around the corner. Whichever company meets those challenges first could well come out on top.
“There is no free lunch in the electronics business, you have to pick your poison, they have chosen one path and we have chosen the other,” Harding said.
If you do not receive this within five minutes, please try and sign in again. If the problem persists, please
email: