By Christian G. Dieseldorff, Industry Research & Statistics Group, SEMI
The DRAM sector experienced a major decline during and following the 2008-2009 financial crisis and eventually contracted — both the number of suppliers and installed fab production capacity. According to the SEMI World Fab Forecast Report, the outlook is now more positive as DRAM bit demand is on the rise and average selling prices improved in both 2013 and 2014. Installed capacity is expected to emerge from negative to positive territory by the end of 2016, but factors for growth are complicated by complex technology issues.
For five years before the economic downturn, yearly growth rates for installed fab capacity trended in high double digits. Looking back to 2007, eleven major companies produced DRAM chips in about 40 facilities globally, with installed capacity growing by 40 to 50 percent year-over-year from 2003 to 2007.
Since that time, the number of companies shrank from 11 players to six, with only 20 facilities in production and three major players (Samsung, Micron and SK Hynix) as the industry consolidated and contracted (see Figure 1 below, from SEMI World Fab Forecast report). Qimonda, Promos and Powerchip left the scene, while Elpida and Rexchip were acquired by Micron. In addition, some front end fabs were converted from DRAM to Logic, Flash or other purposes.
The smaller number of key suppliers has stabilized the DRAM investment cycle and increasingly the manufacturers focus investments on market demand, not on production share gain. Meanwhile, DRAM bit demand is growing for applications such as mobile and infrastructure/servers. One leading memory company has predicted CAGR of 27 percent for bit growth from 2013 to 2017.
Obstacle: New Paradigm is a Loss of Capacity
SEMI’s tracking of fab data reveal that when a company transitions a fab to the next leading edge technology there is a capacity loss. Increased complexity and more process steps mean that these fabs produce fewer wafers per square foot of cleanroom. This trend affects all industry segments, beginning at the 30/28nm node and smaller, and has been observed since 2012. Depending on the age of the fab and product type, this loss can be significant, as much as 10-20 percent (See Figure 2).
SEMI’s World Fab Forecast report tracks nine fabs following this pattern: significant loss of capacity when transitioning to the next leading edge technology node. From 2014 to 2016, existing DRAM fabs are expected to lose a total of about 25,000 wafers per month, every year when transitioning to next leading edge technology node.
To compensate for this and to meet expected bit demand, the industry is beginning to add new capacity with new fabs and lines. By 2015, three or four new fabs or lines will be in operation. Of course, these will require time to ramp up; meaning that net capacity change likely will not shift from negative to positive territory until 2016, when about 3 percent growth is forecast. Figure 3 illustrates how this could potentially affect worldwide DRAM capacity.
The worldwide loss of DRAM capacity from 2010 to 2014 is about 25 percent. The loss of capacity due to technology upgrade kicks in about 2013 timeframe. Before that the loss is due to consolidations, closure and change of product types.
Obstacle: What’s Next, after 15nm?
Shrinking the DRAM nodes has become increasingly difficult. As most companies produce in volume 30nm-25nm, some companies began already to offer 21/20nm node. The next stage beyond that is only just being explored. Will the industry see another shrink down to 1Ynm? Or is this too challenging, and for most, not economically feasible? Other technologies may move forward to eventually replace conventional DRAM, such as non-volatile memories like MRAM (Magnetic RAM), FeRAM (Ferro-electric RAM) and ReRAM (Resistive RAM), and PRAM or PCRAM (Phase-Change RAM). As these technologies surface, DRAM capacity may be challenged again.
In summary, DRAM, the underdog, comes from behind and appears to promise positive growth by 2016. With the introduction of new technologies, it remains to be seen how DRAM capacity will be impacted and how much new wafer capacity will be needed. The SEMI World Fab Forecast Report lists over 40 facilities making DRAM products. Many facilities have major spending for equipment and construction planned for 2015. Learn more at www.semi.org/MarketInfo/FabDatabase and www.youtube.com/user/SEMImktstats.
SEMI World Fab Forecast Report
SEMI’s World Fab Forecast reports lists over 40 facilities making DRAM products. 20 of these are dedicated DRAM facilities and over 30 of these have major spending in 2015 for equipment and construction.
The SEMI World Fab Forecast uses a bottom-up approach methodology, providing high-level summaries and graphs, and in-depth analyses of capital expenditures, capacities, technology and products by fab. Additionally, the database provides forecasts for the next 18 months by quarter. These tools are invaluable for understanding how the semiconductor manufacturing will look in 2014 and 2015, and learning more about capex for construction projects, fab equipping, technology levels, and products.
The SEMI Worldwide Semiconductor Equipment Market Subscription (WWSEMS) data tracks only new equipment for fabs and test and assembly and packaging houses. The SEMI World Fab Forecast and its related Fab Database reports track any equipment needed to ramp fabs, upgrade technology nodes, and expand or change wafer size, including new equipment, used equipment, or in-house equipment. Also check out the Opto/LED Fab Forecast. Learn more about the SEMI fab databases at: www.semi.org/MarketInfo/FabDatabase and www.youtube.com/user/SEMImktstats