Intel (INTC 0.61%) has some of the world's most valuable assets: leading-edge semiconductor manufacturing plants and a world-class technology research and development team. These assets have been utilized to great effect in the company's PC and server microprocessors, and they are likely to prove instrumental in doing battle in the low-power system-on-chip space. But Intel is also gearing up to utilize this to build chips for others. This is quite possibly Intel's most exciting growth area, particularly as it will allow the company to indirectly participate in a number of semiconductor markets that it would not have otherwise been able to.
What type of business is Intel likely to pursue?
Right off the bat, it's important to understand that Intel isn't likely to want to be a general-purpose foundry player in the vein of Taiwan Semiconductor Manufacturing Company (TSM 0.69%) (TSM 0.69%)or Global Foundries; it has repeatedly stated that its foundry ambitions are targeted toward strategic business.
What does that mean? Well, take the recent Altera (NASDAQ: ALTR) deal, for example. Altera designs and markets FPGAs, which is a type of semiconductor product that Intel doesn't build on its own -- nor does it have any intention of building on its own. Further, if Intel got into the FPGA business, it would need to compete with both Altera and its rival Xilinx. Quite frankly, it'd be a mess. So, instead of trying to compete with the established leaders, why not offer one of these two players -- locked in a duopoly and stuck largely on the same manufacturing technology from Taiwan Semiconductor Manufacturing -- a chance to pay a little extra in exchange for a fundamental transistor edge? It's a solid business proposition, but it doesn't end there.
Semi-custom semiconductors
Intel wants to exploit its manufacturing leadership to grow its profitability, but a larger goal is to proliferate Intel's X86 and attendant implementations across a wide variety of end markets. After all, ARM Holdings (ARMH) makes a nice chunk of change licensing IP, so why can't Intel do the same thing? Unlike ARM, though, Intel cannot only license the relevant IP blocks to a system-on-chip vendor -- one that is building chips that don't directly compete with Intel's own products -- but it can offer to fabricate the wafers and then perform the packaging and test. There's plenty of margin to be collected all around.
More importantly, though, this helps the X86 software ecosystem grow. The one thing that ARM has enjoyed is a software ecosystem that has seen incredibly robust growth as its microprocessors continue to see wide adoption in a broad array of end markets. While X86 is certainly here to stay with a very broad and deep developer network, it could be so much more, particularly as the "Internet of things" continues to evolve.
When will it begin to kick in?
It's not likely that revenues from Intel's custom foundry will begin to kick in until late 2014/early 2015 -- and that's at the earliest. While waiting is never fun, it's important to note that this implies some pretty real operating leverage to be had. The spending for these initiatives has been going on -- likely at an accelerated pace -- over the last 2-3 years.
Once the revenue begins to roll in, however, investors are likely to appreciate how quickly the net losses (the spending here is classified under "other," which posted a half-billion net loss in the most recent quarter) begin to evaporate.
The long-term earnings power of this venture isn't clear. But the bigger it gets, the more of the semiconductor sector Intel will own -- and that's a good thing for Intel investors.
