标签: idm

Want even stronger proof that the semiconductor fabless-foundry model is here to stay? Fabless companies grew at a higher rate than IDMs again in 2012, according to market research firm IC Insights Inc. In fact, while fabless company sales increased by 6 per cent last year, IDMs' sales declined by 4 per cent, and total semiconductor sales declined by 2 per cent as a whole. Since 1999, fabless company sales have grown at a higher rate than IDMs in every year but one: 2010, according to IC Insights. That blip on the radar occurred only because the DRAM and NAND flash memory segments—where very few fabless suppliers participate—enjoyed a rare year of explosive growth in 2010. While IDM sales, driven by this growth in memory, increased by 34 per cent that year, fabless company sales growth came in at a still pretty robust 29 per cent, according to IC Insights. From 1999 to 2012, the total semiconductor sales increased at a compound annual growth rate (CAGR) of 5 per cent, according to IC Insights. IDM sales over this period showed a CAGR of 3 per cent, while fabless companies turned in a 16 per cent CAGR, according to the firm.   Back in the day, the big wigs at IDMs like Jerry Sanders of AMD (now a fabless company) famously mocked the fabless-foundry model with the "Real men have fabs" slogan. As recently as last year, no less an authority than Intel manufacturing guru Mark Bohr declared that the fabless model was "collapsing" after TSMC announced it would offer only one flavour of process technology at the 20-nm node. (Although, let's face it, Bohr is not impartial—a growing rivalry with TSMC and other foundries is only one of the many fronts on which Intel is picking fights.) Consolidation leaving fewer players But according to Bill McClean, president of IC Insights, the data shows quite clearly that the fabless-foundry model "is not broken and is not collapsing." While McClean acknowledges that is a decade or so the situation could be different, for the foreseeable future—always a tough thing to define in the semiconductor world—fabless companies, and the foundries who build their chips, are in the driver's seat. What McClean does see, however, is consolidation. Part of the reason for the more attractive growth rate among fabless companies is that there are fewer new ones, and the large successful fabless companies are getting larger and more successful, he said. The number of fabless companies has decreased dramatically since 2000, McClean said. "There aren't a lot of new fabless companies being started, even though China has kind of set their sights on the fabless industry since they never got their IC manufacturing off the ground," McClean said. Consolidation, of course, isn't exclusive to fabless chip companies. IDMs have had their share of consolidation, too. And though a steady stream of chip start-ups—nearly all of them fabless—continues to feed the semiconductor industry, the net of consolidation has left a landscape with fewer players. The most successful of those still standing are getting bigger and more powerful. "In general, even among the IDMs, the big are getting bigger, becoming more top heavy," McClean said.   So, what's the bottom line? It's always dangerous to make long term predictions about the semiconductor industry, which can and does change on a dime sometimes. But recent history has shown the fabless model getting stronger as years go by, showing no signs of a collapse. Will continued migration to more challenging technology nodes alter the equation? Possibly down the road. But don't expect successful fabless companies to transform themselves into IDMs. IC Insights forecasts that by 2017, fabless chip companies will command at least one third of the total chip market—especially if more large companies go fabless, which could well happen. IC Insights' report on the fabless chip companies is part of the 2013 edition of its flagship report, the McClean Report . It can be obtained online through the firm's website .       Dylan McGrath EE Times