Time to change the PC contract DRAM model

 

 

 

Just because you've been doing it the same way for 20 years, doesn't mean it is still right.

The model of the monthly PC DRAM contract negotiations and fulfilment hasn't changed much this century - the volumes are still huge, the spot market remains a key influencer and closing it out every month takes up a lot of time, effort and focus.
But PC sales are close to peak & with DRAM content is effectively static, it is time to do things differently.


Mature industries need to strip out all unnecessary cost in order to make a return. The DRAM companies should make the current model much more efficient.


At present the industry is building tens of millions of memory modules every month, just to support the contract PC market alone. The memory module may have been the predominant way for the DRAM makers to sell their chips 10 years ago but with the PC's share of DRAM output sliding & the Ultrabook rising, it's time for a re-think.


It is not the module itself, rather the support and supply-chain that has built up around it. This requires considerable manufacturing resources. Which have to cope with complex product mixes, rapid turnarounds & increasingly complicated logistics chains. To add insult to injury, you will probably need to wait another 2~3 months after building it to actually get paid for it.


And after all that, we are only talking about a $20 part. Is this really what a cutting edge semiconductor company should be doing?
It's funny, when you consider that whilst most of their PC customers have been doing nothing but outsourcing for the past 10 years, the DRAM companies have remained virtually untouched by this phenomenon.


So imagine instead of building, testing, shipping and invoicing memory modules a DRAM maker decided to partner with someone to do that for them. The efficiencies to be gained in resource, cash-flow and focus by simply selling the wafer to your partner and letting them manage the rest are massive. But it's not just cutting costs in the supply chain that this approach could offer.

So Macho

Monthly contract negotiations between the PC and DRAM makers would still happen directly and of course I wouldn't suggest this model is ready to to applied to the Server module market just yet. But by bringing in a 3rd party into the mix you will change the dynamics of the relationship on PC memory.


The adversarial model we are all used to could evolve into one which delivers genuine consensus in achieving efficiencies & better decisions. This would free up resources, energy and focus which the DRAM makers need to employ elsewhere, as they continue to rush for the PC exits into other more promising sectors.


The PC business has always been a fairly macho environment. The winner takes all mentality (after Intel and Microsoft take their cut of course), was a very efficient way for the industry to grow. But with the industry well into middle age, it's high time that it starts to lower it's testosterone levels.

Please feel free to post your comments below.

If you want to discuss anything in more detail you can email me at memoryman@notsodimm.com


Next Post: Crucial time for Micron.