Guys,
I am in the ASIC business and the Alex Brown report is extremely good. LSI and VLSI are basically dead meat in the general ASIC Arena except in greatly narrowing niches.
By the way I am extremely impressed by your list.
Here's my MACROVIEW:
In general folks are investing in their CORE competencies and vertical integration is disintegrating. 10 years AGO LSI and VLSI had in house CAD tools, designers, and Wafer Fab and Final Test (and outsourced packaging). Today the best CAD tools are from EDA vendors: Synopsys, Cadence and Avanti. The best wafer processing is from folks like TSMC,UMC,Chartered etc.
There is now a large number of fab-less design companies working in conjunction with the foundries in providing ASIC solutions ( I happen to work at one). And I would hazard to say that our people are just as good as LSI's, VLSI's, Cadences etc. In fact as our only leverage is our design expertise and knowledge we (new fab-less companies) tend to attract the best experienced engineers as they can make a more direct contribution, and potentially enjoy significant returns on stock options priced at $0.05 with the hope to go public in a few years at $5-10 apiece.
In terms of design methodologies these are frankly common at most companies and are design tool driven. Every-one basically uses the same design languages (Verilog/VHDL), Synthesis tools(AMBIT/Synopsys) etc.
So the only issue is Marketing/Product strategy.
In the general ASIC business the end customers engineers actually do the system design. For example it is engineers at Sun who actually design the SPARC architecture. Cisco designs it's architecture for ATM/Routers etc.
So in this market segment where one takes the design from a customer and implement it into an ASIC both LSI/VLSI are dead.
Where LSI/VLSI has unique design expertise and IP such as MIPs cores, DVD, etc they have a role to play and a service they can leverage.
So LSI and VLSI are moving/being pushed into more of a niche area where they have clear advantages. The problem is how big are these niches. The major problem is that if a niche gets big enough 5-6 engineers can set up a fab-less design company and start competing with folks like LSI/VLSI. The total CAD budget required is around $1-2M.
Wilf is a very smart guy by the way and so he is trying to move into differing area's such as the Symbios acquisition. Frankly if he could also pick up folks like VLSI/IDTI/LSCC, then he could get to a big enough mass where they can fill Gresham, consolidate and grow rapidly into additional markets - wireless with VLSI, Embedded x86 with IDT/FPGA's with Lattice etc.
Frankly the days of MID-SIZED semiconductor companies (100M-1Bn in sales) with their own fabs is history, you ave to ramp up to $2-3Bn to keep going or you die.
I would expect that LSI will do well long term but after picking up a couple more smaller semi-companies and expanding into additional areas. Short term (next 18 months could be tough).
Regards,
Kash
|
