Memory Gluts: Will the DRAM oligopoly end them?

Summary : TTID: This Time it’s Different or Time-to-Imminent-Disaster?

If you read the business press, it’s become fashionable to predict that the crippling impact of memory gluts are as much history as Polio … oops … Polio’s come back. Meanwhile, we are in the middle of a DRAM glut, prices have collapsed this year, and all these ‘this time it’s different’ (TTID) predictions seem like denial to me.

 

TTID (for which the dual meaning is time-to-imminent-disaster) only work when conditions are not only different, but also only when the difference is significant. So the question is not if it’s different? It’s is it different enough?

 

There is no denying that the market is far more consolidated today.  However, similar predictions were made in 2000 and 2008 … by me in fact. The market did stabilize after 2001. And it did after the financial crisis of 2008 as well. But prices have always stabilized after a short-term shock.

 

 

 

There are several problems with market consolidation as a TTID predictor:

 

  • It is only the supply side of the problem.
    • It ignores significant shifts on the demand-side which, is always at the core of the glut problem.
  • It ignores how capital and technology interacts with supply. What’s happening today has happened before:
    • There has been a significant capital upgrade cycle, which has significantly improved fab productivity
    • That gain is through the significant shrink cycle we are in.
    • Shrink cycles lower cost, raise yield, while raising profits.
    • But, shrink cycles increase output.
      • Which is amplified by increased yields as learning takes place.
    • Output plans typically underestimate these effects.
      • Especially when there’s been a difficult early yield-ramp like over the last two years.
    • Everything is OK, until there is a leftward shift in demand.
  • It assumes that the purely rational homo-economicus is the only decision maker, while also assuming decision makers are only focused on shareholder value.
    • First, shareholder value is a far less important decision factor outside of American public companies.
    • The decision to manage prices in a downturn by cutting supply is fraught with complexity:
      • At the end of an investment cycle, depreciation costs are at their highest.
        • Cutting production raises cost-per-device, which is often a KPI* for management.
      • Cutting back production means loss of scale, which is critical to long-term survival in DRAMs.
      • Maintaining production while building inventory is suicidal for a product that is always depreciating in value.
        • It can mean loss of market share, which is another KPI for management.
        • Politically, it’s far safer to be a victim of the market than to lose market share via your own decisions.
    • Asia plays Go, not Chess.
      • Thus, surrounding your opponent with market share reigns supreme as a KPI.
        • Shareholder value is ephemeral, like killing the king that has many sons.

 

  • Oligopolies always have the problem of cheaters.
    • Even in the case of OPEC, where all the competitors could sit in the same room and agree to control supply, cheating eventually caused it to lose its power. In OPEC’s case, the oligopoly was held together by Saudi Arabia, which had the largest share of output. The effect of systematically losing share eventually got to them.
    • In the case of DRAMs, the era of high concentration has led to companies commonly saying they were holding back on capacity expansion, as they secretly expanded it.
    • Samsung (the largest producer) has often doubled down in capacity just as the market was peaking.

 

Samsung is not Saudi Arabia, because there is no field of DRAMs buried under the sand that is aways there, waiting to be tapped for a buck-a-barrel. Moreover, they are well aware of the fact that the DRAM gluts are like a hemophiliac convention touring a razorblade factory — those that don’t come out bleeding win. They have almost always used gluts to gain significant market share. Moreover, that big DRAM factory they are equipping is proof that this tiger has not changed its stripes.

 

As for DRAM demand, it is weak due to weak markets in PCs, tablets, and smartphones. Moreover, suppliers of tablets and smartphones are tremendous pressure to cut costs as the only untapped markets are in Asia, where price is everything. Moreover, the dominant players are being challenged by upstarts, lessening buyer strength for managing prices from the demand side by managing competitor share.

 

* KPI: Key Performance Indicator

Originally published June 19, 2015 in The Chip Insider.   Author: G Dan Hutcheson

 

You may like this also:

Copyright © 2018 VLSI Research Inc. All rights reserved.