DRAM market-share games shifting
from a knockout to a marathon; 4xnm process and multi-bit/cell as
fundamental criteria to judge NAND Flash production competitiveness
Published Apr.1, 2008
DRAMeXchange
Major DRAM suppliers profited from their aggressive expansion in
2005 and this good fortune continued as the price of DRAM kept on
going up in 2006. However, this boom did not come without
consequences. This rapid growth in capacity, in addition to an
over-optimistic projection on Microsoft Vista's demand for DRAM, has
caused numerous industry suppliers much suffering since 2007. As
those first-tier suppliers, who are finding it difficult to squeeze
second-tier suppliers out, they will see intensified cost pressure.
DRAMeXchange analyst, points out that some DRAM suppliers are still
considering the present DRAM depression as a best opportunity to
expand their market share. The DRAM market-share game is now
shifting from a knockout to a marathon without a winner, and the
whole business loses, according to DRAMeXchange.
Driven by the increasing demand in 2006, the upping of DRAM
pricing motivated major suppliers to budget heavily capex for
expansion. Within less than a year, the sharper-than-expected DRAM
price fell rapidly and eroded profit of many industry suppliers. For
those who lagged behind in terms of process node, a weak cost
structure has led to critical loss.
Pressures from capex and operation, as reflected in shrinking
cash in hands (Cash & Equivalents), prompted some DRAM suppliers to
sell marketable securities for cash, or re-rent their 8-inch fabs
after sales. As in 4Q07, those DRAM suppliers, which originally had
relatively good cost structure, also recorded loss. Given that the
cash in hand at these companies is rapidly declining, even those
suppliers, who originally were supposed to grab more market share,
now feel the pressure and a need to trim their capex in 2008
(instead of aiming at profitability), according to data from
DRAMeXchange.
Despite DRAM suppliers are adjusting their strategy to preserve
profitability, their gross margins are still being seriously eroded
and some DRAM suppliers even found ASP edging to variable cost
level. While industry players generally interpret a capacity scale
back as putting white flag in this DRAM race, none of them has made
such announcement.
DRAMeXchange analyst, points out that some DRAM suppliers are
still considering the present DRAM dilemma as a best opportunity to
expand their market share. Thus, they still plan for consistent
expansion in order to strengthen its production economy of scale and
expel those second-tier suppliers simultaneously. Nevertheless,
their plan is not going as well as they have expected, because they
either postpone their new fab construction, maintenance, trim
headcount cost or announce bankruptcy to prolong their life in this
industry.
If top-tier suppliers' " perfect game theory " encounters
barriers, the nature of competition will shift from a knockout of
second-tier suppliers to a marathon. As those first-tier suppliers
who are finding it difficult to edge second-tier suppliers, they
will see intensified cost pressure. DRAMeXchange believes a better
pricing environment will only be possible if these suppliers slow
down further expansion and adjust capacity, or everyone loses.
4xnm process and multi-bit/cell as fundamental criteria
to judge NAND Flash production competitiveness
Apple started adopting NAND Flash as the storage media for its
iPod lineup since 2005. In 1Q06 and 1Q07, a glut in the NAND Flash
industry was observed. This scenario has repeated in 1Q08, due to a
noticeable sequential drop in shipments in the four key NAND Flash
applications (handset, digital camera, USB Flash drive; UFD and MP3/PMP)
for this quarter, as well as the consistent expanding of 12-inch
wafer capacity.
Although chipmakers can adopt some measures to adjust the
inventory on hand in slowing the price drop magnitude and maintain
their profitability, the core competitiveness very much depends on
the pace of their advance production to 4xnm and 3xnm, and the
development of 3bit/4bit per cell FG (Floating gate) or CT (Charge
Trap) Data Flash. Recalling production roadmaps of leading players
on the tape-out schedule for 4xnm and below process, Toshiba and
SanDisk are taking the lead.
According to news release from Toshiba and SanDisk, the die area
of a 16Gb MLC NAND Flash is only 120 mm2, representing a 30%
decrease in comparison to a 16Gb MLC that is fabricated under 56nm.
The shrink in die area indicates that NAND Flash makers will find
their per unit cost minimize as more chips can be produced on the
same wafer after process shrink. This also translates to a stronger
profitability in return.
Migration to 3bit/4bit per cell means more bits of data can be
stored in the same cell. Although the migration will sacrifice some
performance, per unit cost decrease will also bring a lower
production cost, as evidenced in the lower cost for producing 2bit
per cell MLC NAND Flash than SLC component. We foresee NAND Flash
makers to introduce their respective FG or CT-type NAND Flash on
4xnm and 3xnm over the coming quarters, as well as the introduction
of 3bit/4bit per cell products.
2H March NAND Flash contract price review
The NAND flash contract price of 2H March, 2008, falls at the
range between 0-15% due to the low season, account-close factor and
over supply in the market. The price of SLC products has declined
greater than MLC products due to the time lag after MLC price
dropped for the past few months. With comparatively less supply in
2Q08 and increasing demand, the price is very likely to stop falling
and begin to stabilize.
NAND Flash spot price recap, Mar 24-31
In the SLC segment, 1Gb dropped by 1.7% to US$1.78; 2Gb dropped
by 3% to US$2.60; 4Gb dropped by 9.1% to US$3.69; 8Gb dropped by
7.5% to US$7.31 and 16Gb dropped by 9.8% to US$16.69. In the MLC
segment, 4Gb stayed flat at US$1.90; 8Gb dropped by 2.3% to US$2.52;
16Gb dropped by 2.8% to US$4.91 and 32Gb dropped by 2.6% to
US$10.93.
