On May 15, I finally did something that I had hoped that I'd never do: I put in a market order in to sell the entirety of my stake in chip giant Intel (NASDAQ:INTC). Within an instant, a position that I'd started building in 2012 and added substantially to in 2013 was gone.
Intel is a company that I wanted to own for Warren Buffett's ideal holding period: forever. But, after careful consideration, I decided that this wasn't a stock that I wanted in my portfolio any longer.
Although Intel's financial execution over the last few years has been solid, deeper down the company's technology execution has been quite poor.
To understand what I'm talking about when I say that Intel's technology execution has been poor, I need to explain some basic concepts.
A computer processor, such as the one that's powering the device that you're reading this on, is ultimately judged by the following criteria:
- How fast is it?
- How power efficient is it?
- Does it have a competitive set of features and technologies integrated to support the device maker's vision for what the device can do?
- Is it economical to build?
These characteristics depend on two key factors. First is the manufacturing technology that the chip is built on, which has a great influence on 2 and 4 than it does on 1 and 3. The better the performance, power, and area of a chip manufacturing technology, the better the products built using that technology will be.
The second -- and arguably more important -- factor is the architecture of the chip. A modern computer or data center chip usually has many different pieces of technology built into it, such as processor cores, graphics cores, memory controllers, and so on. The quality of each of these individual technologies, as well as the technologies that connect them all together, has a huge impact on the performance, power efficiency, and capabilities that a chip delivers.
Historically, Intel operated on a so-called tick-tock development model. In a "tick" year, Intel would deliver a substantial manufacturing technology improvement and a modest architectural enhancement. In "tock" years, the company would make minor enhancements to the underlying manufacturing technology but make big changes to the design of the processor. This product development rhythm led to a sustained cadence of leadership products that was hard for others to match.
Beginning in 2016, this tick-tock methodology fell apart. Since Intel couldn't bring its 10-nanometer technology into production for the second half of 2016 as it was supposed to, it rolled out a family of chips known as Kaby Lake. The Kaby Lake processors used the same basic Skylake architecture but were manufactured using a performance-enhanced version of its 14-nanometer technology, known as 14nm+.
Then, in the second half of 2017, Intel introduced two products: Kaby Lake Refresh for notebooks and Coffee Lake for desktops. Kaby Lake Refresh was built using the company's 14nm+ technology while the Coffee Lake chips utilized a further-enhanced 14nm++ technology. Both Kaby Lake Refresh and Coffee Lake increased core counts relative to their predecessors, but they, once again, used the same architecture as Skylake.
Earlier this year, Intel brought Coffee Lake to high-performance notebook computers (Kaby Lake Refresh was targeted at more mainstream products) and later this year, Kaby Lake Refresh will be replaced by a product known as Whiskey Lake. This should, yet again, be based on the company's Skylake architecture.
This lack of architectural innovation on Intel's part isn't because Intel can't build newer, better chip designs, but because its newer chip designs were developed to be manufactured on the company's 10-nanometer technology. Since 10-nanometer tech has been delayed by multiple years, so too has Intel's architectural and product innovation.
Now, I don't want to blame Intel's manufacturing group for the entirety of these problems. Intel's product development groups should have had the foresight to plan around potential problems with its 10-nanometer technology, especially given the significant difficulties the company faced in bringing its 14-nanometer technology to market.
Why I sold Intel
The reason I sold Intel stock is simple: I think the company's product competitiveness as a result of these seemingly endless 10-nanometer delays will take a large hit as Intel's product innovation continues at a snail's pace while competitors move much more quickly because they have access to more capable chip manufacturing partners and technology.
I think that loss of competitiveness will manifest itself across Intel's core businesses in the form of both market segment share loss and gross profit margin erosion as the company tries, potentially in vain, to stem that market segment share loss.
In fact, over the last year or so, Intel has been losing share in markets like gaming desktop processors and premium notebook computers, and has even begun losing share in the data center processor market. I think that share loss could potentially accelerate as a result of Intel's botched execution around its 10-nanometer technology.
Now, I might have been willing to weather that storm if Intel management had made a compelling case to investors as to how it intends to reassert its leadership and, at a minimum, stop losing share (though, ideally, they should have a plan to recapture lost share). But right now, Intel is still publicly maintaining that the delays in its 10-nanometer products and its "architectural innovations" on 14 nanometer (which are practically nonexistent) will allow it to deliver "leadership products." It's almost as if management is oblivious to the share that it has been losing in certain segments and in denial that this share loss could continue into the foreseeable future.
Or, perhaps worse, Intel is fully aware of the potential negative business impact that its 10-nanometer fumbles will have and its corporate communications team is carefully crafting messages that are, on the surface, designed to keep investors in the dark and keep the stock price as high as possible for as long as it can.
In either case, until Intel starts showing dramatically improved product and technology execution, I think the upside in Intel stock is minimal and the potential downside rather substantial. That's a poor risk/reward proposition and, consistent with that view, I sold my Intel stock.