Intel (INTC +2.00%) has signaled plans to launch its next-generation Skylake family of processors at some point during the second half of 2015. Kirk Skaugen, who heads the company's PC Client Group, said Intel will release both desktop and mobile Skylake parts during 2015.
Leaked material (courtesy of CPU World) claims the low-power variants of the Skylake processor -- aimed at slim notebooks and two-in-one devices -- will use LPDDR3. This is the standard memory today in notebooks and other mobile devices, but LPDDR4 will be available by early next year and promises to bring speed and power efficiency improvements.
Given Intel's significant focus on platform-level power efficiency for its PC and convertible products, why wouldn't the company choose the lowest-power, highest-performance memory available at the time?
Let's talk economics
During a presentation at the Intel Developer Forum, management showed the following slide illustrating the expected pricing of DDR4 versus DDR3 for a 16GB RDIMM over time (note that laptops and phones don't use 16GB RDIMMs, but the relative cost is what is important here):
Source: Intel.
The premium of DDR4 to DDR3 is, as shown in the slide, expected to be 25% in 2014, and that premium will come down to about 10% in 2015. While 10% might not seem like much (and it could be lost within a multi-hundred dollar system), this is an extra cost that will lead to one of the following outcomes for 2015 PCs:
- PC vendors will cut corners elsewhere to accommodate the more expensive memory within a fixed price point, potentially hurting the user experience.
- PC vendors will raise prices, which could lead to lower sales and thus reduced processor sales for Intel.
- PC vendors' margins will contract.
None of these outcomes is good for the PC industry, particularly given that the market as a whole is not growing much (if at all). In the current environment, PC vendors either want features that bring sufficient improvement to drive system upgrades at flat-to-up selling prices, or they want to cut costs and take advantage of any price elasticity that exists in the PC market.
Including premium-priced LPDDR4 to save a relatively small amount of power compared to the power drain of the CPU, display, and storage subsystems just doesn't seem like a particularly good idea for most PC vendors. And not surprisingly, "most PC vendors" are the customers for Intel's processors.
Why can the mobile chip vendors make the upgrade?
Qualcomm's (QCOM 0.06%) upcoming Snapdragon 810 system-on-chip supports LPDDR4 memory, which makes perfect sense for a number of reasons. First, the largest customer for Qualcomm's premium processors is likely to be Samsung for premium Galaxy smartphones.
Samsung manufactures its own DRAM, which means it gets LPDDR4 "at cost." I assume a significant part of the premium over LPDDR3 is due simply to the fact that it's a higher-value technology to customers rather than an intrinsic manufacturing cost increase. For Samsung's competitors, such as HTC and LG, to keep up, they'll also need to use the Snapdragon 810, which means buying LPDDR4.
Also, given the much smaller batteries of ultra-mobile devices relative to laptops, any power efficiency gains from LPDDR4 will be more impactful in flagship phones and tablets than in notebooks, where memory is far from the biggest consumer of power.
