Micron Technology (MU) is one of the world's largest memory manufacturers and the only memory manufacturer based in the Western Hemisphere. Micron is in the late stages of an acquisition of Elpida. Elpida is itself a consolidation of Japanese former memory manufacturers NEC, Mitsubishi, and Hitachi. The last obstacle to the acquisition is the expiration of a 30 day period for bondholders to appeal the Japanese court decision to approve the acquisition. That period expires on March 28, 2013.
What will Micron look like after the closing of the Elpida acquisition?
First of all, Elpida will be accretive to Micron earnings. We were told during the recent Micron earnings conference call that Elpida will add 45% to Micron's total 300mm wafer capacity. From recent articles, Elpida's wafer capacity, between two factories, is 180,000 wafer starts per month. So, .45*X=180,000. X=400,000, the number of Micron wafer starts pre Elpida. Since this data came on March 21, 2013, it would be safe to assume the 400,000 includes a January re-negotiation of an Inotera JV that added 60,000 wafers to Micron's total output.
So, we will have a Micron with 580,000 300mm wafer starts per month after the close of the Elpida deal.
During the conference call management also revealed that the Singapore NAND facility had only ramped to 60,000 wafers of its design capacity of 100,000 wafers. This makes the immediate or short term total capacity for Micron of 620,000 wafer starts per month.
One of the important things that this acquisition gives Micron is what management calls "scale". What they mean by this is that Micron now has the ability to shift entire fabs between DRAM and NAND depending on the condition of the market for each technology. In the past, Micron, for example, might have had only two DRAM fabs and therefore would have had to move 50% of DRAM production to NAND if they wanted to take advantage of higher NAND pricing. That would involve going to zero output for the re-configured fab for a few months and, maybe, the "chunk" of DRAM would be a higher percentage than Micron would want to re-configure. Post-acquisition, Micron will have four major DRAM fabs, one of which will soon be in preparation to convert from DRAM to NAND. So, "scale" gives the freedom to make changes in output to take advantage of market conditions. This is a new and important flexibility for Micron.
Now, let's see what can be expected from all these wafers:
Looking at the cost of goods for both Micron and Elpida (from a pro-forma presentation in a form 8 filed on March 5 2013) and dividing by the wafer numbers from above, the average cost of a 300mm processed wafer appears to be about $1800. That number is surprisingly close for both companies.
The worst wafer in the recent past would be a wafer of 2Gb DRAM chips. The chip size of this device is about 40 square millimeters, so a wafer would yield about 1500 chips. The price of the 2Gb chip was $.80 as recently as November 2012. So wafer revenue would be $1200 or $600 below cost of manufacture. That is a negative 50% gross margin for that DRAM chip! The best wafer would be a wafer of 128Gb MLC NAND chips. The chip size of this device is about 190 sq. mm, so a wafer would yield about 300 chips. The price of the 128Gb chip is about $11 in the comparable time frame. So wafer revenue would be about $3300 or a $1500 positive gross profit or 45% GPM.
It doesn't take a genius to figure out that the ability to move from a negative 50% gross margin to a positive 45% gross margin by using "scale" to alter manufacturing output in the direction of the 45% GM business is a good deal. The simple existence (not actually doing it) of the flexibility has a tendency to increase the ridiculous low price of DRAM. And it has done just that, the spot price of DRAM is up well over 100% since December 3, 2012.
Finally, that 7.4 million annual wafer starts at $3300 per wafer would produce $24.4 billion in revenue at 45% gross margin. Give or take what you please on the above numbers. In the event of a general memory shortage (also alluded to during the earnings CC), the price of both DRAM and NAND could well rise above the 45% gross margin level. Prices could also go lower than the 45% GM level, but since there is no new industry capacity coming on-line anytime soon, the case for lower prices is harder to make. Obviously it will take some time for these higher spot prices to become higher general prices.
Micron, compared to about three years ago, has gone from "zero to hero". Three years ago Micron looked like they were spiraling the drain themselves with no chance of becoming the winning memory industry consolidator. In the shadows, Intel (INTC) has always been there when Micron needed some assistance, and they could be there again through the IMFT joint venture if NAND runs too short to service the solid state drives for the PC sector. This brings to mind the Beatles tune, "With A Little Help From My Friends." Nonetheless, Micron management has earned my respect.
Micron, after Elpida, will be the second largest processor of silicon in the world behind only Samsung. Since about $12 billion of Samsung's output is consumed by Samsung internal, Micron will now be the "swing player" in the external merchant market for DRAM and NAND. Micron, in three years, has become the master of their own destiny.
When we get the final word that the waiting period for appeal has expired, I would expect something of a pop in the Micron share price. Buy anything Micron, but a little gamble on April 2013 10 calls might make your weekend a bit better.