Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Semi Equipment Analysis -- Ignore unavailable to you. Want to Upgrade?

To: Elroy who wrote (77040)	9/14/2017 10:38:03 PM
From: Sam	Read Replies (1) \| Respond to of 95757

Are you free lancing for IC Insights or something, Elroy? They keep saying what you keep saying. At least, they acknowledge in the article below that they have been wrong so far and it doesn't look like they will be right any time soon. But of course they keep toeing the same "Its inevitable" line that you repeat. DRAMs: Is a Price War Looming? September 13, 2017 By Gina Roos epsnews.com I'm not going to put the entire piece here. The headline is more dramatic click-bait than the body of the text is, but certainly it frames the bear case and is tilted toward it. Here is the most relevant excerpt: “Previously, when DRAM capacity was tight and suppliers were enjoying record profits, one or more suppliers eventually would break rank and begin adding additional DRAM capacity to capture additional sales and market share. At that time, there were six, eight, or a dozen DRAM suppliers,” said Matas. “If the supplier was equipping an existing fab shell, new capacity could be brought on-line relatively quickly (i.e., six months). A greenfield wafer fab—one constructed on a new site—took about two years to reach high-volume production.” So it begs the question with only three major DRAM suppliers, will the same thing happen? It only takes two to engage in a price war, said Matas. “Our forecast expects it to happen. Look for it in 12 to 18 months after one DRAM supplier decides to add more capacity.” “In true cyclical fashion, one DRAM supplier will decide to add more capacity to meet customer demand. Not wanting to be left behind or lose market share, other DRAM suppliers respond by adding capacity as well,” he added. “Soon, with all of the added capacity, the industry produces more DRAM than OEMs need and inventory rises. DRAM makers offer lower prices in order to sell everything they are producing. System OEMs then gain the upper hand in negotiating to get the lowest possible price and it sets off another price war.” At least Matas acknowledges that it might be different with three suppliers than it was with a dozen. Indeed, I think it is fair to say that it will be dramatically different. Before, you would have 5-10 new fabs; now you will only have three, assuming, as I think it is fair to assume, that each vendor will over time build one new DRAM fab, relying on node transitions for more growth in chip supply to meet the projected increases in demand. But that doesn't take into account that they can bring those fabs up gradually, as they have done with some NAND fabs over the past few years. In the Bad Old Days, they had incentives to bring up new fabs as quickly as possible, but with only three vendors, they have different incentives. They are all expanding their businesses into higher value products than selling chips. That gives them incentives to keep chip prices higher rather than lower, as higher prices will hurt their competition in those other downstream products. And each of the three of them are heavily into 3D NAND and NVM chips, which will limit the amount of investment that they will make in DRAM. Matas talks about them switching NAND capacity into DRAM capacity, but this is nonsense. NAND demand is exploding too, and shows even fewer signs of abating than DRAM does. And the conversion process has become much more expensive, complicated and time consuming than it was with previous generations of DRAM and NAND. Indeed, with 3D NAND, the production process is completely different. In other parts of the article, he agrees that neither Samsung nor Micron have evinced any appetite for a price war or for dramatically expanding greenfield capacity. Indeed, not even Hynix has done that, although they have said that they are building a new DRAM fab. What he doesn't say explicitly but does at least hint at is that while the number of suppliers is down by a lot, the sources of demand are up by a lot. And are varied. As he says in the piece, mobile demand is down but server demand is strongly up. And not just due to numbers of servers, but to how much DRAM each server requires. The price of DRAM has gone up by so much that new greenfield capacity will certainly be needed, especially because 1x and 1y production is very different from 2x production; it reduces the number of wafers that can be produced in the same clean room space. Which means that more clean room space is required to produce the same number of wafers, although since the chips are smaller, each wafer can be cut into more chips. I think that Matas is ignoring important factors that will lead to a more powerful and longer lasting memory chip cycle than anyone has been looking for. It is, in a sense, a good thing that bears like Matas exist, as it implies that there are still plenty of potential buyers out there whose minds can be changed over time if the results continue to be as strong as I think that they will be.