Bookmarked

Added to your bookmarks on the Seeking Alpha homepage

My first SA article on Intel (INTC) was published on October 1, 2012, a little over a year ago. The stock price closed at $21.66 on that date. Intel closed at $24.60 on November 13, 2013, a gain of 13.5%. Not bad I guess, if you are a brain dead mutual fund, but for an options player like me 13.5% is a gigantic "thud".

The original investment thesis on Intel had two aspects; overwhelmingly advanced process technology, and an apparent build-out of an enormous amount of production capacity. That thesis remains intact today.

51,413 people received this article by email alert
Add your email to get alerts on INTC too:

Get email alerts on INTC »

This article was sent to 369,757 people who get the Investing Ideas newsletter. Get the Investing Ideas newsletter »

So where are we today?

I ran across a couple of videos (1, 2) from a pretty neutral source that seem to support the idea that Intel is way ahead of all competitors on process technology for raw performance, but raw performance doesn't solve all problems.

The first video confirms that you really need "gate last" to do finfets (TriGate). TSMC (TSM) will be alright since they decided to follow Intel on gate last technology, but all the rest of the foundries have a problem because they followed IBM (IBM) with "gate first" technology. In the video, Lucian Shifren claims the foundries are missing a number of technologies and much of the knowledge base required to manufacture finfets. He might know something about that since he was with Intel during the development of the TriGate process technology.

This first video (28 minutes) is imminently understandable and should be required watching for any serious Intel investor.

The second video is just interesting and gives a little insight into the complexities and problems of these bleeding edge technologies.

So, the bottom line is that Intel is, indeed, way, way ahead on process technology and the foundries, including Samsung, have some real challenges ahead of them and are unlikely to catch up with Intel.

Now, let's look at manufacturing capacity.

Here are a couple of links (1, 2) to pictures of the latest, greatest Intel fabs. The first link, if you scroll down, is a graphic of fab 42 in Chandler AZ. The existing fab is in the background and is one of the three fab campuses that enable Intel to produce over $40 billion worth of advanced semiconductor products. The building in the foreground is the new fab complex called fab 42. Fab 42 appears to dwarf the older fab.

The second link is a graphic of fab D1X in Hillsboro, Oregon that shows fab D1X Module 1 and Module 2. Mod 2 is being constructed even before Mod 1 is complete. As a matter of fact, Mod 2 was begun in 2013 when the softness in the PC business was a well-known fact. Again, either of the new fab modules dwarfs the existing fab complex, which is one of the other of the three fab campuses that make up that current $40 billion in revenue. Presumably, the third campus is in New Mexico and has been re-fitted recently. There are two other 300mm fabs being re-fitted in Ireland and Israel. The new chandler fab and the Hillsboro fab are scheduled to be equipped for 450mm production.

Here's the deal:

To the best of my knowledge neither of these two monster fabs are on-line today. Judging from the sheer size of these facilities, a rational person could come to the conclusion that either one, fully ramped, could produce all the current fab requirements of Intel at its present size…and shut down all the in-operation current fabs!

Here is an Oregonian article about some slowdown in Intel Chandler in finishing fab 42. In the article is a quotation, "Intel is not using its fab capacity. They have so much more fab capacity than they need right now," said Jim McGregor, president and principal analyst at TIRIAS Research in Arizona.

So, we have support that the Intel technology is still ahead and likely to stay that way and we have independent confirmation that the Intel manufacturing capacity is massively over what is needed for the present business level.

Why so much capacity?

The entire mobile semiconductor business could perhaps consume the output of one of the fab D1X modules. I have racked my brain trying to think of some business opportunity big enough to utilize all the new fab capacity. I keep coming back to NAND flash for the conversion form Hard Disc Drives to Solid State Drives. There simply is no other business that even comes close to requiring the scale of those two new fab complexes.

If there is anything to this speculation, those new fabs would have to be absolute "wafer shooters" and maybe even 450mm wafer shooters, in order to bring the manufacturing cost of 128Gb NAND chips down to the $2.50 range. That price would allow Intel to build 128GB and 256GB SSDs for a cost of about $15 and $30, respectively. Those products will sell for $50 and $100 respectively at 70% gross margin, and still be priced below the best cost of any other potential competitor, including Samsung (OTC:SSNLF).

I don't know how Micron (MU) plays into all this; I suspect that Intel will buy them with stock, at an earnings accretive price, in order to BE in the memory business with both feet.

There is some kind of big stakes game being played out here and I intend to be in it, with a ton of leverage, until it becomes clear what Intel has up its corporate sleeve. Whatever they have in mind has to happen soon because both of those super plants are ready for equipment and an assignment. $10 billion worth of mobile and $10 billion worth of foundry won't come close to filling that capacity. If Intel intends to acquire Micron at a reasonable price, they will have to move before the investing world discovers the true value of Micron.