In the semiconductor industry, the mantra has always been, "cheaper, faster, and smaller." This principle has been the steam engine that evolved the gigantic, room-sized supercomputers in 1998 into today's thin and light laptops. The fundamental building block of all semiconductor products is the transistor, and those with the cheapest and fastest transistors are ultimately poised to win the race.
Let's talk about Intel
It is no secret that Intel (INTC 0.35%) has the world's best transistors when performance, power, and area are taken into account. The company routinely moves to smaller geometries years ahead of its nearest competitors and, at the same time, has made countless innovations at the actual transistor level that have kept manufacturing technology at least a generation ahead on power and performance.
Now, the natural inclination here is to raise the red flag on these claims, particularly given that Intel's product offerings in the mobile world often lag those from competitors in a number of meaningful ways. For example, Intel's 22-nanometer FinFET-based Bay Trail system-on-chip, while offering leadership CPU performance, tends to lag the latest from Qualcomm (QCOM +15.15%) in a number of key areas (feature integration, graphics performance, etc.). How does one reconcile the claims made above with these truths?
It's not just about manufacturing
There is no doubt that Intel has the world's best semiconductor technology, but the actual design of the products (known as the "micro-architecture") is just as important, if not more so, to the performance, power, and cost of a particular chip. Intel's 22-nanometer designs weren't created with density in mind, nor were they intended to have the level of integration demanded by todays' tablet and smartphone markets.
Intel gets it
Well, that's the bad news. The good news is that Intel looks to remedy these deficiencies with two of its upcoming products. The first, a high-end platform known as "Broxton," will be built on the company's next-generation 14-nanometer process, designed with density in mind, and integrate the correct features into the chip commensurate with a leadership tablet/smartphone part in mid-to-late 2015.
At the low end, the company realizes that most mass-market phones (and, increasingly, tablets) will require integration of connectivity (Wi-Fi, Bluetooth, GPS, NFC, etc.) and cellular (either 3G or LTE/LTE-Advanced), as well as the other blocks that typically come with a mobile SOC. Intel will bring its first-generation part, known as SoFIA, to market in late 2014, built on TSMC's (TSM +0.53%) 28-nanometer process. A follow-on part in early 2015 will be similar, but with integrated LTE-Advanced (rather than 3G).
Moving away from TSMC in 2015
In late 2015, Intel will, as it stated at its most recent investor meeting, move SoFIA to its internal 14-nanometer process. Not only does this give Intel a cost-per-transistor savings, but it enables some pretty significant performance and power improvements. Intel will also see higher utilization of its factories (which could improve gross margins), and by the time the 14-nanometer product rolls around, Intel will already be building off of a decent-sized market share base, contra-revenue free.
How do Qualcomm and MediaTek compete?
With Intel currently offering sub-optimal products, Qualcomm and MediaTek essentially "own" the smartphone and tablet apps processor markets, even without in-house manufacturing (or even a manufacturing technology advantage). However, if Intel is out in late 2015 with a fleet of top-to-bottom 14-nanometer, density-optimized processors, then it will be able to take some rather significant share from these competitors.
Qualcomm will be harder to take down than MediaTek, given its gigantic cash hoard and enormous patent licensing stream. Further, Qualcomm drives TSMC exceptionally hard in a bid to get access to next-generation transistor technology. Transistor technology is a critical enabler.
Foolish bottom line
At the 14-nanometer generation, Intel has a chance to take the lead with a significant chunk of the mobile market and defend its server share rather aggressively. The design teams need to deliver, though, using this amazing transistor technology provided by the process teams. If they can, Intel will become like a force of nature in the mobile market -- one that cannot be ignored.
