Comments on Lee Goldberg’s – Death of Telecom Centric skills in Chip industry

I’m catching up on topics that I missed over the summer. Thus …I’m a tad late on this 😳 better late than never

I’m catching up on Lee Goldberg’s “Paving Paradise” post which laments the dearth of semiconductor companies actively involved in the development of TDM-based technologies while these seminconductor companies may be missing an opportunity as the Internet goes thru its current “speed bump” for Video and Andrew@Nyquist’s addition “The Madness of the Carrier Chipset Market” where he says the chip guys are doing the right thing.

I agree with Andrew on this. The chip guys are doing the right thing in ignoring TDM transport. Here are my points.

* 1st Point – Telecom Semi’s never seriously penetrated the SONET/Transport chip market. It has been ASIC all the way. Saying Telecomm Semi achieved ~20% share would be “generous”. The exception would be SerDes devices where Telecomm Semi’s could provide GaAs, BiPolar, or SigE technologies.

* 2nd Point – The re-opening the wounds of forays into the penetration Transport equipment is not something that I relish. PMCS focussed on what I call “selling snow to eskimos” twice ( 88-92 (link to previous article) and 2002 -> ) Each foray innocently began with solid ( but not spectacular) plans. Unfortunately these plans were later hijacked with the following vision, “It’s a large market with untapped potential (ie the ASIC focussed un-washed who would switch to ASSPs). ASSPs should be able to double their share or more.” Then reality re-surfaced ( ie these boxes differentiate on hardware … so we can only win one big customer … oops number 1 ….each major geographic region has a different set of top 3 vendors … oops number 2 … our 1 win only sells in one region … and the litany of problems grows … back to the original plan.). Transport has been the “crack” of Telcomm semi executive. Each foray was “it’ll get better with time … we’ll convert ’em”. I’d like to say that the lack of investment in Transport markets is because Telecomm executives understand that this is “crack”, but I can’t. What they do understand is significantly “shrinking” markets.

* 3rd Point — Most New transport equipment has evolved past TDM services. There could be a point in lack of Transport silicon, but not in TDM. But the reality is that optical is too good for it’s own good. Technical Improvements have only made the revenue pie smaller 😦

* 4th point

Lee G comments … What concerns me as much or more than the actual ability to create an ASIC is the potential loss of the critical knowledge base required to design stuff that really works to telco specs and interoperates properly with the many weird and subtle variants that have evolved within the slop of the SONET/SDH protocol.

This has happened. NT, LU, etc are shadows of themselves in this area. But the new network paradigm doesn’t live and die on TDM services anymore. It doesn’t care about Metro-Ethernet either. It cares about Applications over dumb pipes. It may not matter that TDM folklore is lost, but Lee is probably right in that there will be hiccups.


One thought on “Comments on Lee Goldberg’s – Death of Telecom Centric skills in Chip industry”

Comments are closed.