Ruminations on the electronics industry from David Manners, Senior Components Editor on Electronics Weekly.
GloFo, Samsung Overtake Intel On Mobile Process Technology
Intel has squandered its manufacturing advantage when it comes to mobile.
GloFo and Samsung say they’ll have a 14nm finfet process in mass production at the end of this year. The process will be available to foundry customers and optimised for low power SoC designs. Samsung’s and GloFo’s biggest customers are for mobile SOCs.
Intel goes into mass production on 14nm this year as well, but is not intending to make mobile ICs on 14nm for a year to 18 months.
Asked when he’ll first make a mobile IC on the 14nm process Intel CEO Brain Krzanich says: “We’ll bring that in on our 14nm process either late 2015 or early 2016. We’re still battling back and forth on how fast we can bring it in and at what impacts that has. 14nm is the technology there.”
With Intel bringing its first 14nm non-mobile products out this year that means mobile products are still second class citizens when it comes getting access to Intel’s latest processes.
Yet Intel still talks about its ‘manufacturing advantage’ in respect of mobile ICs when its mobile ICs are still bring made by TSMC – a situation which will continue to ‘late 2015 or early 2016.’
Samsung’s 14nm process will be available in its fabs in Hwaseong, Korea and Austin, Texas. GloFo’s will be available in its fab in Saratoga, New York.
“With this industry-first alignment of 14nm FinFET production capabilities, we can offer greater choice and flexibility to the world’s leading fabless semiconductor companies, while helping the fabless industry to maintain its leadership in the mobile device market,” says GloFo CEO Sanjay Jha.
The GloFo/Samsung 14nm process claims up to 20% higher speed, 35% less power and 15% area scaling over industry 20nm planar technology.
The process claims to be the first FinFET technology in the foundry industry to provide true area scaling from 20nm. The technology features a smaller contacted gate pitch for higher logic packing density and smaller SRAM bitcells to meet the increasing demand for memory content in advanced SoCs, while still leveraging the proven interconnect scheme from 20nm to offer the benefits of FinFET technology with reduced risk and the fastest time-to-market.
PDKs are available now, allowing customers to start designing with models, design rule manuals, and technology files that have been developed based on silicon results from 14nm FinFET test chips.
Mass production for the 14nm FinFET technology will begin at the end of 2014.
So, as GloFo’s Jha says, it looks as if the fabless mobile guys are going to have the jump on Intel.
Intel is going to have to continue paying TSMC for its mobile wafers for another 18 months at least and, on top of that, Intel has to pay its customers to use them. Bernstein Research says Intel is paying customers $50 a shot to use an Intel mobile IC but the total cost to Intel could be much more.
And with the GloFo/Samsung announcement that it will be mass-producing mobile ICs on 14nm this year, Intel’s aspiration that, at 14nm, it will make an integrated LTE baseband and apps processor chip in in its own fabs which will be competitive with Qualcomm, Apple and Samsung looks like a pipedream.
Either Intel can’t develop an advanced mobile process worth tiddly-squat, or its CEO hasn’t got the clout within Intel to insist that its mobile products are made on its latest process.
I wish I knew which it was.