According to DRAMeXchange, a research division of TrendForce, the commodity DRAM yearly output ratio has been on a downtrend due to changes in market demand. This year, total commodity DRAM production fell below 50% for the first time, and the category illustrated the weakest profitability. As for non-commodity DRAM, server and mobile memory has benefitted from the rise of cloud services and the extensive use of handheld devices. TrendForce indicates, server and mobile DRAM output is to account for more than 40% of total DRAM production in 2013, replacing commodity DRAM as the mainstream memory on the supply side.
PC DRAM price began dropping in July 2012. Although the DRAM market is on the road to becoming an oligopoly, some DRAM manufacturers have yet to find outlets for excess capacity. In order to maintain capacity utilization rates, some makers carry on to focus on commodity DRAM production, which has resulted in the severe difference in supply and demand for 2H12. Taiwanese DRAM makers are confirming the greatest losses, as they have limited market share when it comes to the server and mobile memory sectors; the manufacturers are at a far higher risk than other makers due to their high commodity DRAM ratio, and will soon have no choice but to reallocate capacity or back out of the market. TrendForce predicts the industry to see another wave of large-scale capacity cuts in the near future, as there is a gross inconsistency in the cost structures of different makers. Commodity DRAM output is likely to decline below 40% in 2013, while bit shipments to shrink next year and yearly PC DRAM production to drop for the first time ever.
Cloud computing stimulates server builds; content per unit over 50GB in 2013:
Taking a closer look at server DRAM, as it is difficult to break into the market, the bulk of the server memory sector is conquered by top-tier DRAM makers. In terms of revenue, Samsung semiconductors takes over 60% of the market, while SK Hynix and Micron come in second and third, respectively. On the market, numerous module makers are fighting tooth and nail for small to medium business bids; the oligopoly status is even more clear on the sever memory market than in the commodity DRAM sector. Looking at historical pricing, as server product design and packaging is alike to that of PC DRAM, the server and commodity DRAM price trends usually demonstrate positive correlation. Server DRAM price falls when commodity DRAM price falls, but server price declines are usually a step behind commodity price decreases.
Beginning in the 2H12, 8GB modules have been the supply mainstream, as 4Gb chips have gradually matured. The price wars have been intense, with a low of US$40 predicted by the end of this year. As DRAM content per unit is likely to rise from this year’s 43GB to over 50GB in 2013, makers are increasing the production of 16GB modules. 16GB is to likely become the mainstream in the 2H13, at which time its selling price is anticipated to have fallen to the US$60-70 range, representing a major profit compared to commodity DRAM. In light of this, manufacturers are placing priority on server DRAM production in technology migration plans. Makers advance to 20nm processes in 2013, improving cost structure and profitability.
Mobile DRAM remains industry star; demand bit growth at least 50% in 2013:
As for mobile DRAM revenue market share, Samsung still takes the lead with more than 60% of the market, while SK Hynix, Elpida, and Micron are more or less head-to-head in terms of process technology. Competition is harsh on the supply side, resulting in severe drops in average selling price; ASP fell by 10% or more each quarter in 2012. Looking ahead to 2013, smartphone shipments are predicted to watch nearly 30% YoY growth, with a projected minimum of 8.3 million units shipped. With high content per unit growth and the demand contribution from tablets and ultrabooks, demand bit growth to hit 52.7% YoY next year, the highest figure for all DRAM categories. Also in 2013, the mobile DRAM market see increased diversification; LPDDR3 to be the focus of makers’ product development. LPDDR3 not only matches the power efficiency of LPDDR2, but provides a higher data rate and greater bandwidth. TrendForce predicts market acceptance to rise quickly, as LPDDR3 has a broad application area; the memory can be used in ultrabooks as well as smartphones and tablets. LPDDR3 is likely to display the same pricing as its predecessor within three quarters of volume production, thereby becoming the mobile memory mainstream in 2014.
eMCP solutions, which combine mobile DRAM and eMMC, is to also become an important product for DRAM makers to provide if they are to improve their competitive advantage. As eMCP has a cost advantage and shorter client testing times, the product is looked upon favorably on the demand end. By the end of 2013, eMCP is predicted to symbolize over 40% of mobile DRAM products. Currently, the Korean makers are leading the industry in terms of eMCP technology; Samsung’s technology in particular is mature. Mobile DRAM price drop has been less severe than that of other DRAM categories, but as new competitors arrive on the scene next year, price drop to likely intensify in the 2H13. Based on the current market price, LPDDR2 2Gb chip price is nearly three times higher than DDR3 2Gb price; makers are profiting far more from mobile memory than from PC DRAM, which will attract more makers to enter the mobile market, increasing the competition.
Big gets bigger as Samsung steers DRAM industry:
The DRAM landscape is mostly set for 2013. Leading DRAM manufacturers are, in some cases, two generations ahead of other makers in terms of process technology, and the broad dissimilarity in cost structures has resulted in dramatic differences in profitability. As market share leader Samsung takes in nearly half of industry revenue, the DRAM supply chain and price trends to be mostly influenced by Samsung’s strategies.
Samsung recently announced its capex for 2013 would be less than 50% of this year’s figure, obvious indication that even the industry leader holds a conservative outlook on the DRAM market. Looking at technology migration, beginning with the 25nm process, physical barriers as well as equipment limitations – EUV lithography machines are necessary – make continued advancement difficult. Migration to the 2xnm process is to be even more difficult, as higher capex figures and more risk is involved. TrendForce predicts Samsung’s conservative outlook to help the industry’s supply and demand balance in the long term, as the maker shifts its focus from market share to profits. Whether or not DRAM industry revenue and profits to return to healthy levels depend on with reallocation or withdrawal of excess capacity; supply and demand will optimistically see improvement as the industry enters the era of EUV technology