- Intel's node advantage may not last for long.
- Taiwan Semiconductor, Samsung and GlobalFoundries are planning to aggressively shrink their manufacturing processes.
- Intel could witness an increased competition in its mobile, internet of things and server divisions.
Intel (NSDQ:INTC) has dominated the high-performance x86 segment over the past decade singlehandedly. It’s strategic focus on deploying technologically advanced fabrication processes, ahead of its peers, allowed the chipmaker to gain an upper hand over its competition in terms of raw performance and performance per watt metrics. However, Intel’s node advantage may not last for long.
Rival chip manufacturers namely Taiwan Semiconductor, GlobalFoundries and Samsung are upping the ante and are looking to advance their chip fabrication technologies at a rapid pace over the next two years. In fact, there is reason to believe that they might give Intel a run for its money. Let’s understand what’s happening here and how it could impact Intel’s business going forward.
Let me start by saying that Intel has so far been operating on a tick-tock model. While every “tick” cycle represented a node shrink, a subsequent “tock” cycle brought a new architecture. The whole tick-tock cycle took 2 years to complete until very recently. Intel was deploying billions each year to maintain the timeliness of the aforementioned upgrade cycle which, in turn, allowed it to have smaller and more advanced nodes than its competitors. But things have recently changed at the Intel headquarters.
In a bid to reduce capital expenditure, Intel decided recently that it’s coming generations of CPUs would see a tick-tock cycle that would last for three years, up from the current two years. This looks like a smart financial move over the short term period as it would free up cash required for mergers and acquisition related activities, but that’s all the benefit there is. Intel won’t be investing as aggressively as it used to in furthering its technological prowess, which essentially means that competition can finally catchup.
To put things in perspective, Intel has historically enjoyed a 9-12 month long manufacturing lead over its competitors. What this means is: whenever Intel introduced new die shrinks in the past, it typically took 9 to 12 months for its competitors to introduce a similar fabrication node. But after the elongated tick-tock cycle, this lead will narrow and ultimately vanish going forward. If Intel slows down its pace of progression, its competitors are bound to ultimately overtake it over the next 2-3 years.
This is pretty much evident from the publicly available manufacturing roadmaps. For starters, Intel’s 10nm chips are slated for mass-release by Q2 2017. On the other hand, recent reports reveal that it’s rivals namely Taiwan Semiconductor (NYSE:TSM), Samsung and GlobalFoundries are expected to release chips based on the same node during Q1 2017, Q4 2016 and Q1 2017 respectively.
Granted that product roadmaps don’t always perfectly align with actual release dates, but the data available to us does highlight the general trend of Intel’s narrowing manufacturing lead over its competitors. But what does all this mean for Intel’s business and its shareholders?
Implications On Business Segments
There are several ways in which Intel gets negatively impacted due to this narrowing of lead. The technical advantages of shrinking nodes are already explained here so I won’t repeat the same thing, but the crux of it all is that smaller lithography tends to yield faster and more efficient chips. This basically means that by 2018-19, when all fabs will have the same lithography, their chip performance would theoretically be quite similar as the difference in performance will only depend on metrics such as Ion/Ioff and SRAM cell sizes.
Speaking of business impacts, it’s a real possibility that ARM-based chips would find their way into the server industry post 2018 in a meaningful way. With similar theoretical performance levels and clients already looking to move away from x86 due to specialized server needs, enterprises might end up picking customized ARM offerings for their hyperscale computer networks. This would be a big blow to Intel.
The data center division is the most profitable business division for Intel with the segment representing 56% of its total operating profits during FY15. So naturally, chipzilla’s earnings would take a serious hit if it lost market share to ARM-based offerings in its financially most significant business segment.
There’s another issue. We’ve already seen that Intel, even after having a node advantage and much denser chips, has had to provide subsidies to OEMs in order to make its mobile chips appealing to manufacturers. Once the lithography standard becomes uniform across the industry, Intel’s hopes of establishing a foothold in the mobile SoC segment would pretty much be quashed as ARM-based mobile chips would then carry better performance and efficiency metrics over their previous iterations. Intel would have to work even harder in order to snatch a meaningful market share from ARM-based mobile offerings.
What does it mean for Intel shareholders?
Intel may be the current leader in the x86 segment but its lead may not last for long. With its lithography advantage shrinking, and a big question mark on the future of its mobile, server and internet of things business divisions, I believe that Intel stock is poised for a correction. The chipmaker’s stock has historically commanded a market premium for its hegemony and established market position, but since the competition is catching up fast, I believe the premium is no longer justified.