Analyst Romit Shah with Instinet (by way of Barron's) recently offered up some commentary around microprocessor giant Intel's (NASDAQ:INTC) planned higher-than-expected operating expenses for next year.
One possibility that Shah points out is that Intel could be investing heavily in support of a potential contract chip manufacturing deal with Apple (NASDAQ:AAPL).
Intel currently designs and builds the processors that power Apple's Mac computers, but the A-series processors that power its iPhone and iPad product lines are designed by Apple and manufactured by other chipmakers, such as Taiwan Semiconductor Manufacturing Company (NYSE:TSM) and Samsung (NASDAQOTH: SSNLF).
"We have no insight into discussions between Intel and Apple, but within our mosaic, we believe an Apple foundry win would be highly sensitive, which might explain why Intel is being uncharacteristically vague in its communication and waiting until its analyst day in February to provide more details on spending," Shah writes.
Shah also says that "a high-volume foundry win for Intel would materially increase operating expenses." He goes on to add that "part of Intel's value-add as a foundry is that the company takes on design work for the customer."
Here's why I don't think this explanation holds water.
Why Intel is being so vague about 2017 spending
In the past, Intel would issue its full-year financial guidance in January and then go into further detail on the key business and financial details during an analyst day held in the spring. However, former Intel CEO Paul Otellini retired in May 2013, so the company had to push out what would have been an analyst day in May to November 2013. In 2014 and 2015, Intel hosted its analyst days in November. The company would probably have hosted its next analyst day this month, had it not undergone a CFO transition in September.
I suspect that the reason Intel is being so vague at this point is that it usually saves these details for either the end of a given fiscal year or its analyst day, whichever happens to come first.
Would operating expenses really go up a bunch for Intel?
The next point that Shah makes is that if Intel were to build Apple's chips for it, it would also have to take on a significant amount of the design work. To understand what Shah is talking about, it's important to step back and review the two basic development models for fabless-semiconductor companies -- i.e., chip companies that outsource the manufacture of their chips to third-party manufacturers.
There's what is known as the "customer-owned tooling," or COT model, and there's also the application-specific integrated circuit, or ASIC, model. The following passage from EETimes succinctly describes the differences between the two:
"Traditionally, COT companies designed integrated circuits to the physical design level (GDSII) and sent those designs directly to foundries to manufacture the wafers. COT companies then outsourced to assembly, package, and test vendors. An ASIC vendor, on the other hand, provides a turnkey service, managing physical design to delivery of packaged and tested parts."
Intel offers services analogous to the ASIC model, but these days Apple is almost certainly operating under the COT model, meaning that it does the heavy lifting on the chip implementation.
So if Intel and Apple were to strike a chip-manufacturing deal, I doubt that Intel would be implementing Apple's chip design, though there is still an intense amount of collaboration that must happen between chip manufacturer and chip designer.
For some financial perspective, TSMC, a company that supports hundreds of fabless chipmakers, including Apple, incurred total operating expenses -- R&D as well as SG&A -- of $2.79 billion in 2015.
I find it hard to believe that Apple alone could move the operating expense needle for Intel at this point.