Boring Boring DRAM

George would be right at home in the DRAM business these days. 


The DRAM companies have been enjoying a steady increase in monthly contract pricing so far in 2012. But that run came to a halt in June, as Tier 1 PC makers are expected to hold the price flat (avg. $21.50 for 4GB).

Disappointing PC sales in Q2, an increase in their DRAM inventories and less concern about the Elpida situation, has meant the top PC makers are now in a better position to resist the DRAM maker's demands.

Contract prices are likely to remain flat at best in July, after the PC makers have completed their Back to School builds and much of the Windows 8 build ahead. But assuming we see no shocks on the supply-side, then it is likely the contract price may begin to fall after that.

If we do see decreases, don't expect too much though. PC DRAM is in little danger of slipping into a significant over-supply, so contract pricing it is still likely to trade within a narrow band for the rest of the year i.e. 4GB DDR3 $23~$17 (very narrow in the context of DRAM's history).

The spot and after-markets, which has been operating at a near 10% discount to the contract so far this year, are likely to continue to do so in the 2H, or at least until there is any significant change in the Elpida situation. Demand in the local System Integrator and upgrade channels has been abysmal for nearly 9 months now & there is no prospect of relief in the short term.

In the past when DRAM prices where at such low points in the cycle and when they accounted for less than  5% of the PC BOM, this always spurred on MB growth and following that the price rebounded. But nowadays there is no need nor appetite for increasing mainstream PC DRAM content.

It's all a bit of a sorry state of affairs. DRAM used to be the most dynamic sector of the semiconductor business, its 'wild west'. Whereas nowadays it is becoming so predictable and boring it's more 'west country' than 'wild west'.

So DRAM is boring right now, big deal. The interesting thing is though the relative predictability we see on DRAM (and NAND) this year is masking something very big that is coming behind it. The memory business is going thru a serious structural change that will re-shape it & it's participants for the remainder of the decade. Some memory makers have already recognised this and are changing their models successfully, others are paying mere lip service to it, in the hope that the usual cycles will return - most of these guys will be gone by 2020.