There has been quite a lot of focus on Intel's (NASDAQ:INTC) low-power Atom line as the company's main driver into tablets. Not only are the Atoms full system-on-chip products with a lot of specialized, mobile-focused IP blocks (such as a dedicated image signal processor and, in the future, connectivity and cellular), but they're typically designed with cost in mind. While these products will likely constitute the majority of Intel's tablet volume, it's tough to ignore Intel's upcoming chip, known as Broadwell-ULX, for convertibles/expensive tablets.
Haswell-ULX was close, but not quite
Interestingly enough, Intel launched a chip known as Haswell-ULX. This chip was intended for fanless tablets/convertibles and was sold under the Core i3/i5/i7 "Y" series brand. These chips sported a thermal design power "TDP" of 11.5W and a scenario design power "SDP" of 6W. While this was low power enough to end up in 11-inch tablets with a fan or aggressive 13-inch designs without a fan, this was still a ways from being close enough to go into "standard" 7-10-inch class tablets.
The good news is that these chips did find their way into devices like the Sony Vaio Tap 11 which, while not fanless, did offer pretty good battery life and performance in a fairly light form factor (although it was by no means perfect). However, the chip to really watch is the next generation part known as Broadwell-ULX.
Broadwell-ULX looks much more suitable
At the Intel Developer Forum, Intel gave a public demonstration of its next generation Broadwell platform for 2-in-1's and Ultrabooks.
While the immediate change visible in this picture (Broadwell chips on the left, Haswell chips on the right) is that the die size of the processor improves substantially, the not-so-obvious change is the extreme reduction in thermals and power consumption.
Indeed, according to CPU-World, the lowest-power Broadwell parts will sport a TDP of 4.5 watts (this is a high-end smartphone/tablet chip class) and an SDP of just 2.8 watts. This suggests that Intel is seeing a pretty dramatic improvement in power consumption along with the significant die-size reduction (in no small part due to the move to Intel's next generation 14-nanometer manufacturing process).
This could be very interesting
A rough estimate of the die size of the Broadwell CPU suggests that it's about an 80mm^2 die. This is for two very high performance CPU cores, fast I/O, and what is likely to be a pretty hefty integrated graphics for this class of product. If Intel can actually get this into tablets/convertibles, then it could not only command a pretty significant performance lead over the rest of its peers, but it could do so with very favorable economics. If Intel can actually fetch PC chip prices and offer PC chip performance in this small of a die size, then its margin structure should be quite nice.
AMD could struggle further
If Intel is selling small die size, high performance parts into Ultrabooks and high end tablets, then it is likely that Intel's 14-nanometer Atom parts (which will be branded Celeron/Pentium for the low end of the PC market) will be even smaller and cheaper to make. Since AMD (NASDAQ:AMD) may end up using either TSMC's or Global Foundries' 28-nanometer processes likely throughout 2014, it could end up selling larger dies with lower performance/features into the low-end PC market.
Now, AMD bulls argue that AMD can differentiate on graphics prowess, but the problem is that while AMD has been cutting research and development investments, Intel has grown its R&D investments rather substantially.
Does it really stand to reason that Intel hasn't beefed up its graphics capabilities, particularly given its process lead? Does it also stand to reason that AMD can continue to advance while its PC market share continues to decline and its share here dwindles? The strongest argument, of course, is that AMD is still a major player in discrete graphics and that this will be enough to fund future graphics IP R&D. This may ultimately be true, but if Intel combines leadership architecture and leadership manufacturing, then AMD may find its position worsen competitively and, with respect to PC chip ASPs and cost structure, economically over time due to sheer physics.
Foolish bottom line
Intel's 14-nanometer PC chip efforts should be particularly interesting, especially from a platform cost/economic standpoint. Broadwell should be low power enough to make 2-in-1 tablet/PC convertibles viable and sleek. Whether the OEMs actually put out designs that can really do this next-generation platform justice is another point entirely.