This week, EETimes reported that United Microelectronics Corp (UMC) announced that they will skip the 20nm node in order to focus on providing customers with 14nm FinFET tapeouts before the end of 2015. UMC is now the second-largest chip foundry, meaning that this move is important in further establishing the architecture norms for the industry.
Competition turns to smaller
The strategic market event behind this decision is to keep in line with other global chip manufacturers, especially Samsung, which announced back in mid-February the mass production of "the industry's first mobile application processor using the advanced 14-nonometer (nm) FinFET process technology," as they reported in this press release. Samsung, in turn, has been hot on the heels of Taiwan Semiconductor Manufacturing Company (TSMC), who had held a "monopoly in leading-edge services," as the largest chip foundry, as EETimes recently reported.
The competition between these two front-runners has been intense as they compete to offer the best ASP and yield, the most successful architectures, and the leading technology to support demands for increasingly smaller, better performance, lower-power consumption, less heat dissipation, and even faster speeds for multitasking applications – all demands of the growing set of Internet of Things (IoT) components and end-devices alike. Samsung's 14nm FinFET technology has seen good adoption, for example "Globalfoundries is now getting volume production from the 14nm FinFET technology it licensed from Samsung," but the competition continues as TSMC readies their 10nm process to move to production in 2016.
Of course, Samsung and TSMC are not alone in the race to smaller geometries, Intel, always keenly aware of supplying leading-edge technologies to next-generation devices, was among the first foundries to move 14nm into volume production back in August 2014 with the release of Broadwell processors based on this advanced node. We saw many of these new Intel chips and architectures at the Spring tech conferences such as CES, in particular.
Designing at 14nm – partnerships increase
Now that all of the major foundries are in volume production at 14nm, they are seeing increased demand for these nodes from the growth of designs for IoT and mobile SoC applications, especially. Three-dimensional (3D) structures are providing many opportunities for improved designs and meeting the feature requirements as well as improving costs as ASPs drop. Despite these chip geometry opportunities, there are still risks to customers in their architectural shifts down to 14nm. As a result, there is a greater level of interaction between the foundries and the design processes with their customers because of the highly integrated architectures we are moving into.
Certainly, there are important shifts happening at the foundry level. While we may think of these events as being rather far upstream from where many of us engage the global semiconductor supply chain, what happens at the foundry and where their strategies are focused directly affects what happens all the way downstream to the devices available for consumers and what comes back in need of service and repair components. Supply begins at the foundry.