Historically, Intel's (INTC -1.51%) chip fabrication process has been more advanced than its peers'. The chipmaker, for instance, is the first to mass-manufacture computing chips based on a 22nm and 14nm FinFET lithography. This technological lead, in turn, has allowed Intel to repeatedly deliver impressive performance gains under a comparable power draw, before its peers.
Heading further into 2014, however, Intel's hegemony may be challenged. GlobalFoundries and Taiwan Semiconductor (TSM 2.86%) -- leading pure-play chip manufacturers -- are catching up fast. More to the point, their shift to 14nm and 16nm FinFET fabrication process, respectively, may be detrimental to Intel's growth going forward.
The story so far
It's worth noting that, technologically, Intel's highly touted tri-gate, introduced with its Ivy Bridge launch, is essentially the FinFET technology that Global Foundries and Taiwan Semiconductor are currently perfecting. In fact, tri-gate FETs, or field effect transistors, are a subset of FinFETs, originally manufactured as bi-gates.
By adopting a smaller process node, Intel was able to cram more transistors onto the same die. By covering its FETs with a thin silicon coating (which gave birth to the term FinFET), electrical leakages were reduced, thereby making Ivy Bridge chips power-efficient. Consequently, Ivy Bridge chips are about 20% faster and 20% power-efficient than their equivalent Sandy Bridge predecessors.
Intel's 14nm fabrication process, used to manufacture its latest Haswell chips, was believed to have extended its competitive lead. But, earlier this year, GlobalFoundries and Taiwan Semiconductor announced plans to mass-manufacture FinFET chips based on 14nm and 16nm lithography, respectively. This is something that should, in theory, bring all three companies at par, technology-wise.
Taiwan Semiconductor claims that its 16nm process improves the chip's power-efficiency and performance by up to 55% and 35%, respectively, compared to its 28nm chips. GlobalFoundries, on the other hand, claims that its 14nm chips offer a 20% performance improvement and a 35% power-draw reduction, compared to its currently available 20nm chips.
Probable impact on Intel
These performance increments can prove detrimental to Intel's growth going forward. First, Intel is currently the only chip manufacturer to fabricate low-power system-on-chips using FinFET technology. Once GlobalFoundries and Taiwan Semiconductor catch-up, Intel's Atom lineup won't be able to fiercely compete with the companies' latest smartphone and tablet-targeted ARM-based chips, which could hamper Intel's growth in the mobile segment.
Second, several fabless computing chip providers, including Advanced Micro Devices, Texas Instruments, and Qualcomm, are reportedly developing their own iterations of ARM-based 64-bit server CPUs. These tech giants outsource manufacturing to pure-play chip manufacturers like GlobalFoundries and Taiwan Semiconductor.
By using an advanced fabrication process, these companies will now be better-equipped to compete fiercely with the tried-and-tested high-performance Intel Xeon line-up. Intel generates approximately 20% of its revenue from the sale of server chips, and commands a 90% market share in the segment. ARM, however, aims to aggressively enter the server chip segment and hopes to control 10% of the industry by 2017. Naturally, intensifying competition in the space will have a negative impact on Intel's top-line.
Foolish final thoughts
Currently, both GlobalFoundries and Taiwan Semiconductor are perfecting their FinFET technology, with plans to mass-manufacture these chips by early 2015. Whether the chips pose a threat to Intel's dominance will actually depend on their performance and power efficiency improvements.
In the end, the competition will boil down to whose fabrication process is more efficient, rather than which entity manufactures chips using the smallest process nodes. Investors, therefore, might want to keep a close eye on relative performance and efficiency benchmarks, and make informed decisions accordingly.