Sunday, December 3, 2017

Semiconductor sector: quite a ride

Semiconductor stocks have had quite a ride in year 2017. Like a lot of technology stocks out there, also some semiconductor companies start to look expensive. However, not all of the chip stocks have challenging valuations.

Intel core i7-970 pin side
Image: Rainer Kn√§pper, Free Art License

I collected a bunch of well-known large cap chip stocks to the following table and sorted them by forward price per earnings ratio from lowest to highest:

Table. Comparison of selected semiconductor stocks (data source:

It has to be said that this is by no means an apple-to-apple comparison as these companies - while all having to do with semiconductors - are very different from each other. There are manufacturers of memory, central processing units (CPU), graphical processing units (GPU), digital signal procesing units (DSP), field programmable gate arrays (FPGA) etc.

Nvidia seems to be the hottest stock out of the selected ones with forward P/E exceeding 40. Market seems to expect great future for this GPU maker. I guess it has to do with new applications for GPUs like artificial intelligence (AI).

Micron on the other hand has the lowest valuations almost across the board. Only Western Digital has lower price to free cash flow (P/FCF). Memory business has been in the past quite volatile (swinging between oversupply/undersupply) so I guess that may be the reason for the market to take cautios view on e.g. Micron.

Qualcomm seems to have largest pile of cash relative to market gap (P/C), but free cash flow does not look so good. Taiwan Semiconductor Manufacturing Company and NXP Semiconductors N.V have good looking free cash flow (P/FCF). With regards to price per book (P/B) there seems to be many in range of 2-3 while Nvidia is flying far higher than anyone else at 18,77.

Our positions in this sector happen to be the top three companies in the list (lowest forward P/E): Micron, Western Digital and Intel. These companies make the basic building blocks of digital age. While they aren't remotely as hyped as Nvidia is, I believe they too are going to do well as the amount of data that each person generates and consumes is going to grow exponentially.

My first two computers didn't have a hard disk drive. I recall third had something like 30MB. Now "low end" of normal drives hovers around 1TB. That's 25000 times the amount I had in early 90's and likely costs a fraction of what I paid for my first drive back then even if adjusted for inflation. With work memory (RAM) the story is similar.

It does not stop here. Big data applications ("data lakes") in corporate world are already dealing with next level multiples (PB = 1000xTB and EB = 1000xPB). Eventually also individual consumers deal with petabytes (PB) worth of data.

Think VR. I don't think it really takes off until VR gear resolution is on human eye level. What we now know as TV may eventually change to a 360 field-of-view VR gear kind device allowing you to look freely around. That means huge amounts of graphics data to be stored and processed.

AI also needs massive datasets at the time when it's "taught" ("machine learning", "deep learning"). Use of AI is also going to enable analysis of vast amounts of data (think massive realtime automated video surveillance of major cities).

I often think of gold rush and how we may now be in same kind of situation with many technologies (like AI and VR). I do not really think about the people who went out to try to find gold, but the ones who figured out what to sell to those people.

No comments:

Post a Comment