memory chip pricehike

Bad memories

Samsung looks to extend the pain of its competitors as the memory market takes a tumble

There are no better canaries of the health of the semiconductor industry than the memory sector. It is the one that most closely resembles a classic commodity like wheat, pork bellies or orange juice. And it is quite poorly indeed: the indicators are flashing sell, sell, sell. If only the chipmakers sitting on stocks of unwanted memory chips could do the same.

Well, they will in the not too distant future. Unlike those other commodities, they don’t go off all that quickly, though if a crash comes ahead of a transition in process nodes, the weaker suppliers can find themselves having to dispose of older, less dense memory at bargain prices. And that could be a big issue for any that isn’t one of the top behemoths.

Malcolm Penn, president of analyst firm Future Horizons, made the point last year as the downturn got under way that some of these companies would be better off dumping a chunk of their stock in the river or, in these times, at least sending the materials for recycling. There are some precious and rare metals in there, albeit in small quantities, that make it worth mining the wafers and packages.

Though it is one of many downturns the memory sector has seen since its creation 60-odd years ago, some observers see it as unusual. “The chip industry thought that suppliers were going to have better control,” Avril Wu, senior research vice president at TrendForce, told Bloomberg. “This downturn has proved everybody was wrong.”

Many of the prior crashes in the memory market were the result of overbuilding and over-optimism as to the prospects of the technology sector. But, since the financial crash of 2008, it’s been hard for any semiconductor company to get ahead of itself and go mad building too many fabs. Most companies across the semiconductor sector overall have become quite skilled at assessing inventory levels, or least as good as you can expect when customers trying buying through multiple avenues at once when supplies get tight, as they did in the immediate aftermath of the Covid crash. 

But there is only so much you can do when your production lead times are not that far removed from those that farmers face when planting crops. By the time you are sure the demand is there, it’s too late. You should have made that decision months ago. If only you had a time machine. 

The recession we are seeing now is partly due to manufacturers anticipating strong demand holding up when it was in fact weakening. But would you want to be the fab manager arguing for a slowdown in production or a shift to something else as the downturn starts and then discover it was just a temporary hiccup? Those lost sales would hang heavy. 

There is a point at which the best planning goes awry and this is one of those times, as it is largely a demand-led recession not one based on over-aggressive expansion plans. So, this one is a bit different but not that much. It might be the worst downturn since 2008 but it’s not all that different in overall complexion.

Where it does differ is with Samsung’s plans. Whereas other suppliers are cutting capital expenditure to show to the financial community they are making savings, the Korean giant seems to have decided it has seen this movie before and wants to exploit its deep pockets. It is arguably a point at which Samsung can steamroller a few weaker competitors out of the way. The company is not cutting back investment in fabs. As Penn often points out, they take ages to build and fit out. You generally want to be doing that investment as early as you can, or dare, in order to catch any recovery.

It is a risky move on the one hand, as the strategy relies on a return to normal levels of economic activity within a year or two. Then again, if that recovery does not take place, the chip business – and indeed the rest of the world economy – has bigger problems on its plate. If Samsung can exploit its competitors' pain by extending the time during which prices are below profitable levels simply by maintaining output and take advantage of the regular advances in density made possible by deploying newer equipment, then the decision to keep the cap-ex budget where it is will look to be wise a few years out.

On the other side, the strategy could be very painful indeed for fellow Korean producer SK Hynix, which is still trying to absorb Intel’s non-volatile memory operation. Further consolidation seems inevitable. And there are not that many players left to consolidate, which may trouble antitrust legislators in the years to come. However, while China receives much of the attention in terms of international politics, Samsung will be safe for a while.

There could be a lot of pain to go around, as when memory starts  dropping so do other parts of the chipmaking business. Those too could be looking at further consolidation.

Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.

Recent articles