Wednesday, January 07, 2009

Top 20 predictions for semis in 2009

Maybe rather than yet another blog post of 2008 review and 2009 predictions, I'll just flag provocative ones like's Top 20 predictions for semis in 2009 by Mark LaPedus.

Some of my favorite points:

  • EDA's four large vendors will more than likely survive the year intact. ... Despite Cadence's play for Mentor last year, consolidation among the big four is unlikely because of too much product line overlap.
  • The idea that TSMC might buy Cadence (or another large EDA vendor) has been floating around for years. But now that Cadence is on the ropes, is it more realistic? Offering a suite of EDA tools would make TSMC even more attractive to customers. But could TSMC charge as much for the tools as a standalone vendor? Not likely. ... But what if a company from an entirely different space made a play for Cadence just to get its foot in the door in EDA? What if a software company like Autodesk or Oracle decided to make a move? ... It makes for intriguing speculation, and it's not impossible. But it's a long shot at best. After all, who would want to get into the EDA business?
  • [Under "Foundry Fools"] In 2009, I also see Samsung emerging as a foundry powerhouse, eventually pitting TSMC vs. Samsung as the leader in the arena.
  • In ASICs, IBM Micro will scale back its efforts to few accounts. The same with TI. The big ASIC houses--NEC, Toshiba, and others--will spill more red ink. And the fabless guys--eASIC, eSilicon, Open-Silicon and others--will need to consolidate.
  • [Under 450-mm conspiracy] Got some bad news for you Intel: AMD's foundry spin-off, The Foundry Company, is backed by Abu Dhabi. Cash-rich Abu Dhabi can already afford to build 20 450-mm fabs--and fund the equipment vendors as well. So, the joke is on Intel and the other backers of the 450-mm fab--TSMC and Samsung.
  • Except for ARM, semi IP vendors will continue to lose money. Look for a wave of mergers in IP. How about a merger between MIPS and Virage?
  • In the top-10 chip rankings in 2009, 5 out of the top 10 will end up being fabless. Qualcomm, Broadcom, MediaTek, Nvidia and perhaps another will be in the top 10. Global Unichip is an up and comer. The old line IDMs are sinking.

That's a lot of quoting, but there's even more good stuff in the article. The author definitely "names names" of which companies may fail and which will prevail.

1 comment:

Anonymous said...

No way. MIPS blew all their cash on
the Chipidea debacle (bought for
$150mill, then wrote off $100mill).

How long is it gonna take for people
to realize analog IP doesn't scale.
Barcelona design flops. Mosys sells
its analog IP group, etc.