Hey, are Micron and SKHynix just not dirt cheap enough for you? If you want (more) exposure to DRAM, but with an interesting twist, I present to you Nanya Technology (2408-TW).
Nanya is a small Taiwanese DRAM producer -- only about 55K 300mm wafer starts per month (WSPM). Nanya is 87% owned by the gigantic Formosa Plastics Group. In turn, Nanya owns 24% of another Taiwanese DRAM producer, Inotera. The Inotera stake is worth $74.6b at current market prices, so Nanya’s enterprise value adjusted for Inotera is $127.4b. At that EV, Nanya trades for 6X TTM FCF, this is compared to MU at 10.4X (also adjusted for MU’s Inotera stake). If you think Inotera is cheap at current prices, then that makes Nanya even cheaper.
The twist to the Nanya story comes from the fact that Toshiba and SanDisk do not have internal DRAM production. This is an issue for them because inside a phone, DRAM and NAND are packaged together in an embedded multi-chip package (eMCP) to save space, power, etc… The other NAND producers, Samsung, Hynix, and MU, also produce DRAM, so they have an assured supply of both to package to together. Making sure Toshiba/SNDK have a supply of DRAM is an important task. In fact, SNDK employs people to make sure that happens.
January 2014 – Present (1 year 1 month)San Francisco Bay Area
•Manage DRAM suppliers, provide monthly forecast, negotiate price and ensure on-time delivery.
•Coordinate internal and external team to solve supply quality issues
•Manage DRAM supply wisely to meet operational requirement and corporate goals
Still it can’t be that comforting to rely on your competitors for your DRAM supply. And although no one believes it, DRAM is in tight supply. When Samsung dumped a bunch of NAND that it couldn’t use in its phones on the market and blew up SNDK's Q4, you know what they didn’t have excess to dump? DRAM. MU and Hynix guided DRAM bits down ~10% and 5%, respectively, due to lost capacity from node shrinks.
SNDK has partnered in the past for mobile DRAM with Qimonda to put in MCPs. In a recycling of dance partners, Inotera was originally a joint venture between Nanya and Qimonda. Micron purchased its Inotera stake during Qimonda’s bankruptcy.
I suppose this begs the question: why don’t Toshiba and SNDK build their own DRAM production? Putting aside the cost, there is a lot of IP involved in advanced DRAM manufacturing that keeps out new entrants. Nanya, however, has the right to license MU’s 20nm DRAM process. I think that’s fairly valuable asset and the DigiTimes article mentioned that Toshiba’s potential investment would help finance the migration to 20nm. Nanya has stated it will make its decision on how to fund 20nm in Q1 2015.
Putting aside the Toshiba/SNDK angle, which may or may materialize, Nanya is cheap on its own merits. Of course, you have to buy into the DRAM-is-a-permanently-changed-industry thesis, which the market isn’t really buying right now. Many pixels have been spilled on the MU thread on this subject, so I won’t rehash here (we can in the comments if people want to discuss). If you’re really worried about the DRAM thesis, then short Inotera against a Nanya long and you sort of isolate the Toshiba/SNDK angle.
Nanya specifics. As I mentioned, they are a small DRAM producer, only 55K WSPM and 2014 revenue was ~US$ $1.6b. ~80% of their capacity is now 30nm with the rest on 42nm. Obviously, this makes them the marginal producer versus everyone else at 20/25nm. Things will be fine if Samsung doesn’t blow up DRAM pricing and not fine if Samsung does.
2014 was a great year for the company. Their 30nm mix flipped over from 35% in Q4’13 to 70% in Q1’14 and EBIT margins expanded by ~2400bps. Their mix of DRAM output has been changing for the better also, with PC accounting for 26% of the revenue vs. 50% of revenue in 2012. The remainder of their revenue come from better markets such as mobile/low-power (12%) and consumer/specialty (62%).
2015 should be another good year for the company. Bit shipments were actually down 8.6% in 2014, but they have guided to “mid-teens” bit growth for 2015 as their 30nm process matures. They will also move another 10% of their wafer starts to 30nm. The board is also discussing what Nanya’s dividend policy should be in Q1’15.
I do not hold a position with the issuer such as employment, directorship, or consultancy. I and/or others I advise do not hold a material investment in the issuer's securities.
Some sort of transaction with Toshiba or SNDK in Q1.