LONDON – Intel's 22-nm manufacturing process technology with FinFET transistors will be responsible for 25 percent of Intel's IC shipments in the second quarter of 2012, according to Stacy Smith, Intel chief financial officer, speaking on a conference call to discuss the company's first quarter financial results.
Intel is hitting high volume as the company manufactures Ivy Bridge processors for desktop and notebook computers and this is despite the fact that Intel has yet to formally launch the processor. Ivy Bridge is a shrink of the previous Sandy Bridge processor from a 32-nm process to the 22-nm process.
This lines up with the fact that Intel usually does not launch processors until after its first tier customers have designed the processors in, received parts in volume and are ready to launch their own products. A key outlet for a dual-core version of Ivy Bridge will be in so-called ultrabook computers which Paul Otellini, Intel CEO, said will combine the convenience of a tablet computer with the performance of a notebook. He added that 21 models of ultrabook computer are already shipping and that more than 100 additional models are expected to start shipping before the end of 2012.
Intel has been manufacturing ICs on the 22-nm process for several quarters, according to one source and is reported to have started manufacturing Ivy Bridge "in volume" in the third quarter of 2011. One potential reason for holding back shipments of Ivy Bridge until recently was a build up of inventory of the predecessor Sandy Bridge processor, due to slow computer sales in the second half of 2011, which was in turn due to a shortage of hard disk drives.
Ottelini told analysts on a conference call that Intel now has three wafer fabs ramping the 22-nm manufacturing process in the second quarter of 2012 and a fourth wafer fab is due to ramp production in the second half of 2012. He did not identify the particular fabs.
On the conference call Stacy Smith, Intel CFO, said: "One of the tactical benefits we see of being an integrated device manufacturing house is the speed at which we can ramp these factories. We'll have 25 percent of shipment volume in Q2 coming from Ivy Bridge on 22-nm." Smith added that the ramp of 22-nm had been faster than the ramp of Intel's 32-nm process.
When asked about extending Intel's chip manufacturing lead into a foundry business model, Ottelini was non-committal: "You've seen some announcements to that effect, where we've signed up some companies for that activity over the next several years." Otellini said these were being done as a learning experience but that it is being done for profit. Otellini added: "A lot of the work we're doing here is to build the libraries, the tools that allow us as a company, and our designers, to use quick time to market, derivative capabilities, modularity for our SoC businesses going forward. And the best ways to test those [things] is to have some third-party customers to really validate how good they are or where they need some work."
When asked if Intel could take on foundry work for Apple or Qualcomm, notwithstanding that this might compete with Intel's aspirations for its Atom microprocessors, Otellini said nothing was impossible and ultimately such things depended on commercial price points. "From the kind of taste it would leave in my mouth, the Apple win would be a lot more attractive than the Qualcomm win," Otellini said.
Judging by the rise in demand for consumer electronics and mobile devices, the demand for Intel's chips will not go down soon. There will be need for more and more powerful chips that consume less power, mainly for the mobile devices. The consumers send the demand signal, and it sends engineers working towards the next, more powerful processor.
Mary - http://www.jensenmarinedirect.com
Benchmarks for Sandy Bridge vs. Ivy Bridge are coming out:
The power advantage over 32 nm is not dramatic. It is quite scary, maybe it means without the big process change, it could have been much worse.
But it also might mean Sandy Bridge and Ivy Bridge will continue to cannibalize each other to some degree during this year.
One word of caution. Stacy Smith said 25 percent of volume shipments in Q2 would be on 22-nm, NOT 25 percent of production.
Reading between the lines is a dangerous thing to do but it would seem that Intel was producing 22-nm Ivy Bridge in 4Q11 and 1Q12 but had so many stockpiled 32-nm Sandy Bridge chips because of disk drive problems out of Thailand that is decide to ship those first and delay shipping Ivy Bridge.
This could be reponsible for a 22-nm shipment spike in Q2 while production continues to build more gradually.
They are clearly getting a head start on ramping this node. It begs the question as to how much competitive edge is being lost by the companies that abandoned their internal fabs to use foundries. For my personal experience working in a company that did it, the decision was not driven by the engineers but by the managment who were primarily focused on the short term financials.
I think they've been ironing out some 22nm issues too. The TDPs are coming in higher than expected and they are running hot when people try to overclock. They actually can't surpass Sandy Bridge overclocks.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.