Intel’s diplomatically made announcements last week can only be read one way: “We’re bringing our best foundry technology to compete in the smartphone and tablet market.”
Up until now, Intel has brought a technology knife to a smartphone gunfight. But that’s all it cared to bring; the company has not to date put its best-in-class silicon technology on the line against the ARM-based competition. As a result, Intel has little respect among some Wall Street analysts and the stock languishes this week in spite of a record-breaking fourth quarter.
Intel’s current Atom “Pine Trail” generation of systems on a chip (SOCs) is fabricated on circa-2008 45 nm technology that was no longer needed when 32 nm Core processors started shipping in late 2009. Pine Trail is not very competitive against 40 nm ARM processors and is better suited in netbooks than smartphones and tablets. This year’s Medfield generation Atom is being fabricated in 32 nm silicon technology that may offer 11-day standby and 8 hours of Internet browsing time. Better, but the ARM arms race (pun intended) is headed towards 28 nm this year. And Intel’s 32 nm process was rolled out in volume in Q1-2010. In short, Atom until now was made deliberately on last-generation, depreciated fabrication equipment. That changes going forward.
Here’s how Intel gets much more competitive in the smartphone space: spend over $16 billion in capital expenses over three years to greatly expand its latest generation manufacturing capacity. CEO Paul Otellini said at Intel’s January 13th earnings call, “As we have done for decades in the traditional computing markets, we will apply the world’s most advanced silicon transistor technology to these new segments to deliver the lowest power, highest performance, lowest-cost products on the planet.”
Last year, the company announced plans to spend $6-$8 billion on R&D and advanced development for 16 nm and subsequent shrinks in four U.S. fabrication plants. Recently, the company said it would pump $2.7 billion into Israel (where the new Sandy Bridge processors were designed) and is restarting a fab in Ireland at a cost of $500 million. Intel also said they are projecting capex of about $9 billion in 2011, with a lot of that going toward building and equipping incremental high volume manufacturing factory at 22-nanometer.
The next billion Internet clients will largely come from a growing global market in smartphones and new gadgets like tablets. By building factory capacity to produce smartphone-capable Atom chips alongside next-generation Core and Xeon processors, Intel will bring it world-class manufacturing capabilities to compete against the ARM-based rest-of-market. this capacity won’t come online all at once. But by 2014, one hundred years after the guns of August, there will be one helluva market battle. Don’t count Intel out, as many are today. The big, market-moving battles are still in the future.
For those handicapping this year’s battles for marketshare, note two points:
- The global sources of silicon manufacturing have been dramatically reduced by consolidation over the past five years. For soft fabs with ARM licenses, there are fewer manufacturing choices. Intel has capacity for its Atom chips.
- The silicon fabs are filling up as the global economy improves. Giant TSMC is booking Q3 already. A great ARM design that cannot be manufactured quickly is doomed by the inexorable march of technology. Intel is upping the ante in the foundry arms race.