Generally speaking, when Intel (INTC +2.59%) launches a new chip family, it doesn't release all the variants aimed at all the different segments at once. It will stagger the launches, with a complete rollout of all the chips based on a given processor core stretching out well over a year. Intel CEO Brian Krzanich emphasized this point during company's second-quarter conference call.
Oddly, Intel is releasing its Broadwell chips in reverse order from rollouts of prior years. Previously, Intel would luanch quad-core desktop and high-performance mobile processors first, and lower-power chips would follow some number of months later. With Broadwell, we got Core M in late 2014, followed by the 15-watt/28-watt Core i3/i5/i7 processors in early January 2015.
Source: Intel.
The slide above shows Intel plans to launch the higher power (>= 45 watt) and performance chips by "mid-2015." Investors might ask, then, why Intel is releasing the products in this order. This article will explore that question.
Myriad factors
In considering this matter, it's worth stepping back to look at the PC industry today. Laptops and convertibles (particularly thin and light devices) outsell desktops by a significant margin, and in those markets power efficiency is highly prized. Additionally, the first wave of Broadwell chips (Core M and the 15-watt Ultrabook processors) feature relatively smaller die sizes than what high-performance laptop and desktop chips will ultimately feature.
Based on PC sales numbers alone, it makes sense to release the most power-efficient parts first, since that is where the system volumes are. In the desktop and high-performance laptop markets, the 22-nanometer Haswell processors are still best-in-class, and the benefit of moving from Haswell to Broadwell in these systems (which often use discrete GPUs and are often used docked/plugged in) wouldn't be as large as moving the lower-power ones first.
The "relatively smaller die sizes" is also an important consideration. Firstly, smaller dies are generally easier to yield than larger dies, so if Intel still needs to improve its yields, it would make sense to release the smaller dies first. Intel could then apply what it learned from developing those easier-to-build parts to parts that are more challenging.
Let's talk capacity
I reached out to Intel spokesman Chuck Mulloy, who said Intel's Arizona 14-nanometer factory is ramping up now, while the factory in Leixlip, Ireland, should start ramping up later in the first half of this year. This suggests Intel might not have enough 14-nanometer capacity online yet to support high-volume launches of all of its 14-nanometer chips simultaneously.
When the Ireland and Arizona factories are going full blast, though, I expect most of the company's products will move to 14-nanometer. In fact, Intel has indicated that full launches of its next-generation 14-nanometer Skylake parts for desktops and notebooks should occur by the end of this year. Intel has even gone into production on its 14-nanometer tablet part code-named Cherry Trail, and has signaled that even low-cost PC parts will move to 14-nanometer by the second half of 2015.
