In my article The Retina Display Rush, I debated how the race to meet Apple's (AAPL) retina display specs, now that Apple has extended its use to the new iPad, would probably lead to two consequences:
- The increased obsolescence of e-paper devices, and;
- The need for better mobile GPUs. This was met by Apple with the use of the A5X CPU, armed with a quad-core GPU, and will have to be followed by any competitors that choose to use higher-resolution displays.
Upon further reflection, and helped by an extraordinary Anandtech article reviewing the new iPad, I've also come upon two further insights:
- The new SoCs (Systems on Chip, including CPU and GPU) could likely have a much larger area, to accommodate further GPU power to process the much increased number of pixels in the display. This would entail a higher usage of silicon wafers, with possible positive - and much needed - consequences for companies such as MEMC Electronic Materials (WFR). Or;
- This will represent the need for further process node steps to miniaturize the chips further. Although this is ongoing in the semiconductor industry, the fact that the mobile chips are getting so large represents a powerful incentive to go to the next step as soon as possible. The A5X at 45nm process already has an area of 163mm2 - larger than some Intel Sandy Bridge SoCs. I recommend reading this particular page on Anantech's article. The increased need to go to the next process node, especially on such a high-volume industry as smartphones and tablets, should present the need for increased investment in more advanced semiconductor production equipment, with vast implications for that industry - including its heavyweights Applied Materials (AMAT), Novellus Systems (NVLS) and KLA-Tencor (KLAC).
The adoption of retina display in tablets by Apple, and the rush by its competitors to meet those same specs, will imply the need for much more powerful GPUs on mobile devices. These much more powerful GPUs will imply SoCs with either have much larger areas or alternatively, be done at smaller process nodes. Larger areas imply more silicon wafer usage, whereas smaller process nodes will imply higher investment in more advanced semiconductor production equipment.
I believe that to shorten time to market, the first reaction will be towards larger silicon areas much like Apple did, followed by a push towards smaller process nodes, with the investment needed to make this push - starting right now. The trend towards larger silicon areas might not be useful if negated by the present solar sector implosion (due to reduced subsidization in many countries), so only the trend towards higher (cyclical) investment in semiconductor equipment might be useful in terms of being a tradable long idea.