Intel's latest Broadwell-EP server chips. Image source: Intel.
Back in April, well-known stock market commentator Jim Cramer floated the idea that microprocessor giant Intel (INTC 4.99%) should split into two. The idea, according to Cramer, would be to split the fast-growth businesses (data center group, Internet of Things, and non-volatile memory) into one company and the relatively slow-growing PC business as another.
Here's why this idea doesn't make any sense.
Useless replication of R&D spending
The fundamental problem with Intel splitting its businesses in this fashion is that these so-called "high growth" segments of Intel's business depend on technologies developed for the company's PC business. In some cases, Intel will literally take products that were developed as PC processors and rebadge them as Xeon processors to address certain server workloads.
In most cases, though, what happens is that the fundamental technologies that Intel builds to address the roughly $30 billion and nearly 300 million-unit PC market, such as CPU cores, will be used as the fundamental building blocks for the company's server chips.
The chips that Intel sells under its Internet of Things group are mostly processors developed for the personal computer market, though it sells some server-grade chips as well.
If Intel were to split into two, these development expenses would have to be duplicated -- a destruction of shareholder value for both "fast growth" Intel and "slow growth" Intel.
What about wafer scale?
It is generally thought that the fact that Intel manufactures its own chips is a competitive advantage. However, in order for it to make sense for a company to continue to manufacture its own chips, it needs to bring a lot of volume to the table in order to keep those factories utilized.
In terms of total raw unit and wafer scale, Intel's PC business dwarves its data center business. In fact, it's unlikely that Intel's data center and Internet of Things businesses combined could even keep a single leading-edge factory fully utilized today.
So, if Intel were to split into two, the slower-growing part would likely inherit the manufacturing plants and would have to fund the research and development associated with actually making viable process recipes to build the chips.
Would the "fast growing" Intel be contractually obliged to use "slow growing" Intel's manufacturing plants exclusively in some sort of exclusive wafer supply agreement? Or would it have the freedom to use multiple contract manufacturers?
Perhaps in the case of a split, both Intel parts would just ditch its manufacturing capabilities altogether and rely on Taiwan Semiconductor Manufacturing Co. and/or Samsung for their manufacturing needs.
It's just a bad suggestion
The bottom line is that splitting Intel up into multiple parts would be an extremely bad idea. Both "parts" of the "new" Intel would suffer as they lose operating expense leverage and manufacturing scale. In this case, I am confident that the sum of the profits of both companies would be lower than the current profit levels of the one Intel.
