I strongly recommended Micron (NASDAQ:MU) in early 2013 before it was the Wall Street darling that it is now. I and many other people made a lot of money on that call.
I closed out my position after a huge gain, but it was not simple profit taking. My outlook for the company had changed. I knew the company had some major make up to do on DRAM, its most important product line. On average, the company was nearly a technology node behind the market leader Samsung (OTC:SSNLF), but I figured the general equilibrium in the DRAM market would buy them time to get competitive on cost.
I have to confess to being wrong by not understanding Micron's position in the NAND business. I believed that the 16nm NAND would be the cost leader until 3D NAND became established. I was wrong. I didn't understand the strength of triple-level NAND in solid state drives. The weakness of that business at Micron was pointed out by the company CEO Mark Durcan about a quarter ago in such stark terms that I had to reassess my opinion of both the NAND business and the quality of Micron's high level management at the same time. Durcan felt the NAND recovery could take as long as seven quarters.
Once the Elpida acquisition was completed, Micron became a much more definable company. For example, for the three complete quarters since the completion of the acquisition, Micron's sales have been essentially flat at $4 billion per quarter, give or take. No apparent price increases, no efficiencies, $4 billion, period. The company should get to approximately 20% net after taxes which gives $3 per share earnings per year, give or take, and the prospects for anything above that are not clear to me.
I have since moved on to concentrate on Intel (NASDAQ:INTC), which has turned out to be the right move for me. I continue to keep an eye on Micron - one shouldn't turn his back on a company that has made so many people so much money.
Today, I've been thinking about two recent news items concerning DRAM and, therefore, Micron. Both of these pieces are by DigiTimes, whose record for accuracy is somewhat "scattered."
This one is about the well chewed information about a DRAM production bump at Samsung's fab 17.
From this I can thumb up two scenarios, 180 degrees out of phase with each other.
The anatomy of a memory price war
One supplier lowers price on memory (doesn't matter what kind of memory) and takes a piece of business from an existing supplier. The losing supplier now suddenly has an excess of product without a home. That supplier is very likely to cut the price to find a home for the orphaned product (starts to look like the corn market). Remember, Samsung is, on average, about a node ahead of Micron which means that Samsung can sell DRAM for Micron's cost and still make 30%-40% gross margins.
Other customers get wind of the new low prices on the street and expect to participate. The pressure mounts at other customers to lower prices. If Micron does lower prices, Samsung will meet them. And thus, the downward spiral begins.
Micron will be at 20nm within a year, so it will be able to compete better with Samsung. However, in the process of getting to 20nm, Micron will be producing about 50% more DRAM bits, which will also need a home. They will unload any excess on the spot market at whatever price is available. Once started these downward price spirals can take on life of their own and become unstoppable, oligopoly or no oligopoly. We need to remember in very early days of semiconductor memory there were only a few suppliers and that fact didn't stop them from gutting each other in the marketplace.
Today, Samsung is strong relative to Micron in DRAM, it could not be blamed for snatching some market share while it can.
The DigiTimes piece says the Samsung/Apple move is price motivated, which is ALWAYS the beginning of a price war. Nothing said about supply shortages.
On the other hand…
Today, Apple consumes something on the order of a million wafers of DRAM per year - nearly half of the current output of the Elpida operation. All the iPhones and every iPad that I looked at uses 1GB of DRAM. The rumor is that the iPhone 6 will use 2Gb of DRAM, and presumably all the iPads would then follow. That one move would take Apple's mobile DRAM consumption to nearly 2 million wafers.
A million wafers of short-term new mDRAM demand paints the whole market a different color.
This would mean that any business Apple has given to Samsung is supply driven and not price motivated. In that case, you can replace the price war scenario with a serious DRAM shortage scenario with attendant dramatic price increases.
What to do?
Very simple. Watch the iPhone 6 product release on September 9 for the DRAM content of the iPhone 6. If it is 2GB, buy Micron option with both hands AND feet. I will probably be in that crowd.
If the iPhone stays at 1GB, Apple is jerking Micron around for a lower price and you can expect a memory price war of some magnitude, so taking profits in Micron could be the right move.
I've watched DRAMeXchange for the past three weeks and DRAM spot prices have been down ever so slightly (5%), but consistently - pretty much inconclusive, but keep an eye on the spot price for big moves from now to September 9.
Disclosure: The author is long INTC.
The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.