Micron (NASDAQ:MU) reported earnings yesterday afternoon. They were $.85 per share non-GAAP. Not what I had expected, but a significant beat on the street number of $.76, or $.61, or whatever the schizoid analysts had averaged it out to be. They also produced a small beat on revenue. While I would like to have seen a buck a share, the numbers were quite good considering the quarter was one of the two seasonally weak quarters.
After hours, the stock rose, declined, and then rose slightly again, as though the market couldn't figure out whether the numbers were good, bad, or neutral. The same confusion seems to continue today, with the stock up $.85.
So, my call of "Buy Hard and Hold Through Earnings" on March 30 didn't hurt anyone, although I was expecting a much more clearly defined breakout, and we might still get that in the next few days.
$.85 on a seasonally weak quarter would surely indicate about $3.50 per share for FY 2014, which I might point out, is the number that David Einhorn trotted out on CNBC as his rationale for buying a rather large chunk of Micron stock.
The real story here is cash flow, as so eloquently explained by Electric Phred in his article of last night. $5 billion of EBITDA is a pretty nice number for a $16 billion company that is only in the early stages of optimizing operations after swallowing Elpida, a company that has helped Micron double sales in a 12-month period. I love Phred, but he likes to pound on management anytime he can. Personally, I wonder how they can even stay awake for a conference call after what must have been 18 months of crazy activity, the end result of which is an American-based memory company that is a business and technology leader in the semiconductor industry. This group of, what some might call, ragtag old semiconductor street fighters, has survived every government-sponsored competitive challenge of the past 35 years. Not bad guys, my hat is off to you.
Micron doesn't really give guidance as we have come to understand earnings guidance. They publish a sheet of best guesses on the next quarter's bit growth, ASP, and cost per bit. Anyone who can produce a future earnings number from that data is someone I want to have lunch with.
This data is all on a bit basis, so, for example, say a transition is taking place with 2 Gb chips moving to 4 Gb chips. Typically, the 4 Gb chip will sell for less than two 2 Gb chips, so the per-bit price would be lower. Since memory has been and always will be about increasing density, bit ASP will always, or most of the time, decline. The cost of manufacturing one 4 Gb chip is so much lower than the cost of manufacturing two 2 Gb chips that Micron will gladly give up a few pico-pennies to move customers to the much more profitable 4 Gb chip.
Given the above paragraph, the ASP/bit on DRAM can always be expected to move down. It is a remarkable event when DRAM ASP is forecast to be flat. The Q1 forecast for Q2 DRAM ASP was flat. ASP for DRAM actually came in a -1%, still great. Cost for DRAM bits was forecast to be down 11%, it came in at down 8%. Not bad. Bit growth was forecast to be flat, and it came in flat. Now that little tidbit of information is a bit of a distortion, since a rather large DRAM fab was in the process of being converted to NAND. Bit growth of the other fabs must have been quite large.
So, you see what I mean about the difficulty of forecasting off these guidance sheets? The forecast sheets vary from right on LUCKY to so far off that you wonder why they publish them.
BTW, given the emphasis management put on the "tightness" of DRAM availability, DRAM ASP will surprise to the upside, trust me on that.
NAND bit growth was forecast to be up 17%, it came in at 35%. 35% in one quarter! Again, because of the fab 7 conversion. The forecast for q3 NAND is ~-8% bit growth, ASP to be down 3%-4%, and cost to be flat. The management explanation for the decrease in NAND bits was that more NAND would be used in SSDs. SSDs have a longer manufacturing cycle time due to testing and burn in, etc. That extra time will soak up the 8% of NAND as SSD work-in-process. I hope I see the day when all Micron NAND is used internally for higher-level, higher-margin product (SSDs, memory cards, etc.), and none is shipped as components.
Micron is holding off on sampling the 3D NAND chip. Reason being that the 16nm is more than competitive right now and the 3-D chip is slated to be used in system level (SSDs) products within Micron, so there is no benefit in letting competitors look at their product and process. So, SSDs will suck up most NAND production in the future. Good for margins.
NAND nodes: 60% 20nm HKMG, 32% 25nm, 8% something else, probably some legacy 34nm, and a bit of 16nm.
Doug Freedman asked a question about NAND quality, which gave management an opportunity to tell the world that the feedback from customers is that the Micron SSDs are the highest-quality in the market. Neat. Was that a plant? Remember, Micron is the only manufacturer fabbing NAND with an HKMG process.
The Micron numbers are good, and will get better. Apparently, they are swallowing Elpida with very few difficulties. The company is positioning for the inevitable tsunami of NAND-based SSD business. NAND/SSD guarantees high growth well into the future. The company isn't perfect, but they seem to know what needs to be done to get nearer to perfect. At $3.50 per share earnings, the share price is certainly worth $35 today and substantially more in the future.
Disclosure: I am long MU. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.